Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous polyposis patients

Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous... To prevent duodenal and ampullary cancer in familial adenomatous polyposis (FAP) patients, a diagnosis of high grade dysplasia (HGD) plays an important role in the clinical management. Previous research showed that FAP patients are both over- and undertreated after a misdiagnosis of HGD, indicating unwarranted variation. We aimed to investigate the labora- tory variation in dysplasia grading of duodenal adenomas and explore possible explanations for this variation. We included data from all Dutch pathology laboratories between 1991 and 2020 by retrieving histology reports from upper endoscopy specimens of FAP patients from the Dutch nationwide pathology databank (PALGA). Laboratory variation was investigated by comparing standardized proportions of HGD. To describe the degree of variation between the laboratories a factor score was calculated. A funnel plot was used to identify outliers. A total of 3050 specimens from 25 laboratories were included in the final analyses. The mean observed HGD proportion was 9.4%. The top three HGD-diagnosing laboratories diagnosed HGD 3.9 times more often than the lowest three laboratories, even after correcting for case-mix. No outliers were identified. Moderate laboratory variation was found in HGD diagnoses of duodenal tissue of FAP patients after adjusting for case-mix. Despite the fact that no outliers were observed, there may well be room for quality improvement. Concentration of these patients in expertise centers may decrease variation. To further reduce unwarranted variation, we recommend (inter)national guidelines to become more uniform in their recommendations regarding duodenal tissue sampling and consequences of HGD diagnoses. Keywords Pathology · Gastroenterology · Familial adenomatous polyposis (FAP) · Duodenal polyposis Introduction formation of ≥ 100 synchronous polyps distributed through- out the gastrointestinal tract [3, 4]. The standardization of Familial adenomatous polyposis (FAP) affects one in 10,000 colonic screening at a young age and subsequent preven- individuals, which makes it the second most common inher- tive surgery has largely reduced mortality from CRC [5]. ited colorectal cancer (CRC) syndrome [1, 2]. A mutation In addition, individuals with FAP have an increased risk of in the adenomatous polyposis coli (APC) gene leads to the duodenal and ampullary cancer, with relative risks of 331 and 124, respectively, compared to the general population. Consequently, duodenal and ampullary cancer are nowadays * E. Soons the most common cause of cancer-related death in FAP [6, elsa.soons@radboudumc.nl 7]. The histological grade of dysplasia in duodenal adeno- Department of Gastroenterology and Hepatology, Radboud mas plays an important role in the decision-making process Institute for Health Sciences, Radboud University Medical Center, Nijmegen, The Netherlands to prevent duodenal and/or ampullary cancer in two ways. First, international guidelines recommend starting duodenal Department of Pathology, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, surveillance at the age of 25–35 years [8–11]. The surveil- Nijmegen, The Netherlands lance interval is traditionally determined by the Spigelman Department of IQ Healthcare, Radboud Institute for Health classification, which includes histological grading as one Sciences, Radboud University Medical Center, Nijmegen, of four decisive parameters [12]. A diagnosis of high grade The Netherlands Vol.:(0123456789) 1 3 E. Soons et al. dysplasia (HGD) will lead to a shortened surveillance inter- pathology reports from Dutch pathology laboratories, with val in many cases. Secondly, the presence of HGD is a rela- nationwide coverage since 1991 [20]. All PALGA data tive indication for an endoscopic or surgical intervention as are pseudonymized by a trusted third party, securing that it is considered a risk factor for developing duodenal cancer in the PALGA database no personally identifiable data are [9, 11, 13–15]. collected. Data from patients who refuse their data to be As a result of the need to screen for dysplasia in duodenal used for scientific research are excluded from the PALGA tissue in FAP patients, duodenal tissue is routinely seen by database. The scientific and privacy committee of PALGA pathologists. Yet, there are indications that misdiagnoses approved the protocol of this study (Reference Number: of dysplasia occur, which may have major clinical conse- 2020-41). Non-identifiable data makes this study to be quences. For instance, Sourouille et al. described the histo- exempted from ethical approval. pathological diagnoses of 52 duodenal specimens collected We identified all reports with one or more diagnoses of after radical surgical treatment [14]. Surgery was performed duodenal adenoma between 1991 and 2020 from patients in FAP patients with endoscopically untreatable duodenal diagnosed with FAP or who had a prior (sub)total colectomy, polyposis and/or an ampullary adenoma or Spigelman score assuming they were also FAP-patients. If there were multiple IV with HGD at 2 successive endoscopic assessments three records per patients, all reports were included. months apart and confirmed by two independent patholo- gists. They found that in eleven patients (21.2%) surgery was Laboratory, patient and specimen selection performed too late (i.e., cancer was already present), while in fourteen patients (29.7%) surgery was performed too early Based on the search criteria described above, a total of (i.e., no HGD or cancer was present). Both scenarios indi- 5782 reports from 1217 patients and 49 laboratories were cate inappropriate care, as the misdiagnosis of HGD led to identified. To account for small sample variations, reports both over- and undertreatment of duodenal polyposis. from laboratories with < 30 reports (n = 266) or without an Measuring institutional variation in clinical care may help HGD diagnosis (n = 33) over the total inclusion period were identifying inappropriate care and/or suboptimal quality of excluded from further analysis (see Fig. 1). To aim for uni- care and may provide target points for quality improvement formity in our dataset, pathology reports were also excluded [16]. Variation in clinical care is an issue, especially if it if they were inconclusive on the degree of dysplasia (n = 109) is unwarranted. The latter may occur in clinical care when patients undergo care which is not indicated, as illustrated in the above-mentioned study by Sourouille et al. [17]. So far, no studies on laboratory variation in the grading of dysplasia in duodenal tissue have been published. In clinical practice, the guideline that is used for grading duodenal dysplasia is the same as for colorectal dysplasia [18]. In the latter, considerable interlaboratory variation has been reported in grading of colonic dysplasia, with 35% of laboratories reporting a significantly lower or higher frequency of HGD than average; however this information is not yet known for duodenal dysplasia grading [19]. If the same is true for duo- denal tissue, there is need to reduce this variation to prevent over- and undertreatment of duodenal polyposis. The aim of the current study was to investigate the extent of laboratory variation in the histological dysplasia grading of duodenal adenomas from patients with FAP in a nation- wide cohort and, if present, to identify possible explanations for this variation. Methods Data extraction All data were extracted from PALGA, the Dutch nation- Fig. 1 Flowchart representing reasons for exclusion for excluded wide pathology databank. PALGA contains excerpts of all reports 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… or origin of the tissue (n = 2101), revised cases (n = 7), from Laboratory variation in proportion HGD diagnosis patients < 18 years of age (n = 11) or from patients who pre- viously underwent duodenal resection (n = 137). Further- Laboratories were compared by indirect standardization of more, reports of resection specimens were excluded (n = 68) HGD proportions. First, the observed proportion of HGD when information on the number, size and location of the diagnoses per laboratory was assessed. Second, expected duodenal specimens remained unclear in these cases. To proportions of HGD diagnoses per laboratory were calcu- correct for multiple paired measurements, we included one lated with a multilevel logistic regression model. This model specimen per report. This was either the (first) specimen accounts for age, year of the report, number of specimens diagnosed with HGD, or, in absence of a HGD diagnosis, per report, localization in the duodenum, localization at the first specimen that was described in the report. the papilla major, morphology and method used to obtain The final dataset used for further analyses therefore the tissue (i.e., polypectomy or other). Localization in the consisted of 3050 specimens from 926 patients in 25 duodenum at the papilla major, or the method of obtaining laboratories. the tissue were significant predictors in univariate analysis. Nevertheless, as previous studies showed the significance of Data collection these variables, they were included in the final model regard- less [14, 19]. See Supplementary Table 1 for the contribu- For each laboratory, we registered the type of laboratory tion of each variable to the final model. Third, the observed (academic or general), number of patients and number of HGD proportions were divided by the expected (adjusted) specimens. For each patient we extracted age, sex and total HGD proportions per laboratory. This led to an observed/ number of specimens. Only the first report per patient was expected-ratio (O/E-ratio) indicating less HGD diagno- used to describe these characteristics since multiple reports ses than expected when < 1, or more HGD diagnoses than per patient were included for most patients (55.5%). For each expected when > 1. Fourth, by multiplying the O/E-ratio specimen extracted, the year of the histology report, total with the overall HGD proportion, the standardized HGD number of removed specimens per report, degree of dyspla- proportion was calculated. sia (no dysplasia, LGD, HGD or carcinoma), morphological To quantify the amount of interlaboratory variation in type (tubular, tubulovillous or villous), localization in the HGD diagnoses, a factor score was calculated. Factor scores duodenum from first to fourth part (D1-D4), tissue from the are used to illustrate by which factor the highest score differs duodenal papilla or not and tissue obtained by biopsy or not, from the lowest score. The factor score was calculated by was recorded. dividing the mean proportion of the highest three labora- Clinical characteristics were extracted manually. To tories by the mean proportion of the lowest 3 laboratories. validate the extracted data, 10% of the reports were double A variation of factor two is considered to be modest [21]. checked by three additional investigators (ES, MvK and TB). Multiple imputation was performed for the variable mor- Funnel plots phology, which had 26.3% missing data. All other variables were complete. Funnel plots were constructed to detect outliers. In short, these are frequently used control charts in which an out- Statistical analysis come measure for a unit of analysis (e.g., a histopathologic laboratory) is plotted against a measure for the laboratory Descriptive analysis size (“precision”). The O/E-ratios were plotted against their expected values and control limits (95% and 99.8%) were Overall laboratory characteristics were described with included around the target value (O/E = 1). O/E-ratios of respect to laboratory type and number of patients and speci- laboratories outside the control limits are considered outli- mens. Overall patient characteristics were described with ers and perform significantly different from the target value. respect to age, sex and number of specimens. Specimen Additionally, laboratories that lie between both control limits characteristics were described with respect to year of his- can be considered as random variation. tology report, total number of removed specimens per report, location in duodenum, degree of dysplasia, morphology type Explaining and understanding laboratory variation and method used to obtain the tissue. Categorical variables in proportion HGD diagnosis are presented as frequencies and percentages. Continuous variables are reported as means ± standard deviation (SD) To study whether case-mix at least partly explained the or medians (interquartile range (IQR)), in case of a skewed laboratory variation, we compared factor scores based on distribution. the standardized proportions with the factors scores of 1 3 E. Soons et al. unstandardized proportions. Any explanatory influence of observed HGD proportion was 9.4%. The highest stand- our case-mix variables should result in lower factor scores. ardized HGD proportion was 14.9%, whereas the lowest In addition, to further explain the variation, laboratories was 3.5%. All academic laboratories reported more HGD with low and high standardized HGD proportions were diagnoses than average. The mean highest 3/lowest 3 factor compared for laboratory type, number of reports, previous score for the standardized HGD proportions was 3.9, which assessment by another laboratory (i.e., if a different labo- indicates that tissue diagnosed in the highest 3 diagnosing ratory previously assessed tissue of the same patient) and laboratories had a 3.9 times higher likelihood of being diag- degree of dysplasia. For this, the three laboratories with the nosed as HGD than tissue diagnosed in a laboratory from the lowest three standardized HGD proportions and the three lowest 3 laboratories. laboratories with highest three standardized HGD propor- Figure  3 presents a funnel plot showing the variation tions were selected. between laboratories. The O/E (i.e., standardized) ratio is presented on the y-axis, and “expected”, the number of Sensitivity analysis expected HGD-cases per laboratory, on the x-axis. The O/E- ratios varied from 0.4 to 1.6. One laboratory (i.e., 4% of all When the Spigelman classification was introduced in 1989, laboratories) were located outside the 95% control limits. the dysplasia grading was originally graded as mild, moder- Nonetheless, all laboratories fell within the 99.8% control ate or severe, which was changed into a two-tiered system limits, according to what was expected. (low-grade dysplasia (LGD) and HGD) to decrease interob- server variability [18]. To study the effect of this change on Explaining and understanding laboratory variation our data, we performed a sensitivity analysis in which only in proportion of duodenal HGD diagnoses data between 2000 and 2020 were included, after the intro- duction of the two-tiered system. Laboratory variation and To investigate the effect of case-mix adjustment on the extent magnitude of variation were calculated as described above. of variation, we compared factor scores based on standard- Analyses were performed with R version 1.3.1073 (R ized proportions with the factors scores of unstandardized Foundation for Statistical Computing, Vienna, Austria. proportions. The mean factor score for the unstandardized URL: http:// www.R- proje ct. org/) and IBM SPSS Statistics HGD proportions was 7.8, which decreased to 3.9 after version 25 (SPSS Inc., Chicago, IL, USA). case-mix adjustment. The lower factor scores for standard- ized proportions indicate that our case-mix only partially (3.9/7.8 = 50%) could explain the observed (unstandardized) Results laboratory variation. To further identify explanations for variation, Table  2 Lab, patient and specimen characteristics shows direct comparisons between laboratories with low standardized HGD proportions (lowest 3 laboratories) and Table 1 shows the laboratory, patient and specimen char- laboratories with high standardized HGD proportions (high- acteristics. Eight of 25 (32.0%) included laboratories were est 3 laboratories). The lowest 3 laboratories were all general academic. Mean number of patients and specimens per laboratories, while the two of the three highest laborato- laboratory was 42 (range 11–127) and 122 (range 27–748), ries were academic laboratories. In the lowest 3 laborato- respectively. Mean age of patients at the time of their first ries a mean of 70.7 reports (range 55–91) were included, report was 55.7 (± 18.3) years, and 43.3% of patients was whereas in the highest 3 laboratories a mean of 101.7 female. A median of 3 (IQR 5) specimens per patient was (range 55–165) reports were included. HGD was diagnosed included in the analyses. Most specimens (95.5%) were his- approximately seven times more frequently in the highest 3 tologically reported from 2001 to 2020. In more than half of laboratories compared to the lowest 3 laboratories (14.0% reports (53.7%), one specimen was described. Most speci- vs 1.9%, respectively). Both highest and lowest 3 laborato- mens (84.2%) came from the descending duodenum (D2), ries particularly diagnosed tissue from patients who had not with a minority (14.4%) located at the papilla major. HGD yet been assessed by another laboratory (93.1% vs 94.8%, was diagnosed in 9.4% of the specimens. Morphology was respectively). described as tubular in 63.7%. Only 15.6% of specimens were obtained by polypectomy. Sensitivity analysis Laboratory variation in proportion of HGD diagnosis Sensitivity analysis did not reveal significant differences Figure  2 shows the standardized HGD proportions per in highest 3/lowest 3 factor scores (4.25 for standardized laboratory as well as the overall mean. The overall mean proportions), which indicates that the change in dysplasia 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… Table 1 Laboratory, patient and Laboratory characteristics N = 25 specimen characteristics Academic laboratory, n (%) 8 (32.0) Patients, mean (min–max) 42 (11–217) Specimens, mean (min–max) 122 (27–748) Patient characteristics N = 926 Age (y), mean ± SD 55.7 (18.3) Sex (female), n (%) 402 (43.3) Number of specimens, median (IQR) 3 (5) Specimen characteristics N = 3050 Year of histology report, n (%)  1991–2000 137 (4.5)  2001–2020 2913 (95.5) Removed specimens per report, n (%)  1 1639 (53.7)  2 698 (22.9)  3 314 (10.3)  4 158 (5.2)  ≥ 5 241 (7.9) Location, n (%)  D1 232 (7.6)  D2 2569 (84.2)  D3/4 249 (8.2) Localization at papilla, n (%) 440 (14.4) Degree of dysplasia, n (%) No dysplasia 172 (5.6)  LGD 2580 (84.6)  HGD 287 (9.4)  CA 11 (0.4) Morphology type, n (%)  Tubular 1431 (63.7)  Tubulo-villous 717 (31.9)  Villous 100 (4.4)  Removed by polypectomy, n (%) 477 (15.6) N number; min minimum; max maximum; y years; SD standard deviation; IQR interquartile range; LGD low grade dysplasia; HGD high grade dysplasia; D1 duodenal bulb, D2 descending duodenum, D3 inferior duodenum, D4 ascending duodenum; for subsequent analysis only one specimen per report was selected grading in 2000 did not significantly influence our data. See HGD diagnosing laboratories also had higher volumes com- Supplementary Table 2 for all factor scores. pared to the lowest three HGD diagnosing laboratories. Perspective Discussion Our results showed an observed HGD-proportion of 9.4%. This study in a Dutch nationwide cohort shows that there In addition, HGD was diagnosed in 7.0% and 10.7% reports is moderate laboratory variation in scoring HGD in duode- from general versus academic laboratories, respectively. Pre- nal adenomas of FAP patients, as indicated by standardized vious literature on the prevalence of HGD in duodenal pol- HGD proportions. Additional explanatory analyses showed yps in FAP patients is scarce. However, two recent studies that the case-mix of this study could explain approximately briefly discussed this. First, Sourrouille et al. reported that half of the observed laboratory variation. The highest three 3.9% of their included cases were diagnosed with HGD after the first upper endoscopy. Five and 10-year rates of HGD 1 3 E. Soons et al. were 12.1% and 20.8%, respectively [14]. Second, Roos et al. reported that 4% and 17% of endoscopically removed polyps from the duodenum and papilla were diagnosed as HGD, respectively [22]. Both studies were performed in tertiary centers. Therefore, our results (both overall HGD-proportion as well as the HGD-proportion in academic laboratories) fell within the previously reported range of HGD prevalences. To our knowledge, this is the first study to report the labo- ratory variation in HGD scoring of duodenal adenomas in patients with FAP. It is important to realize that the same guideline is currently used for the grading of dysplasia for both colorectal and duodenal tissue [18]. A Dutch study by Kuijpers et al. showed considerable laboratory variability in dysplasia grading of colorectal adenomas, as illustrated by the fact that 13 of 37 (35%) included laboratories were aber- rant, i.e. they reported a significantly lower or higher fre- quency of HGD in colorectal adenomas than average, even after correcting for case-mix. Most of these aberrant labo- ratories (9/13, 69.2%) reported more HGD than expected, which is in line with our results as standardized proportions of HGD were higher than average in 19/25 (76.0%) labora- tories. The effect of volume differences per laboratory was not analyzed in this study [19]. Our results show that the highest 3 laboratories graded more polyps than the lowest 3 Fig. 2 Bar chart representing standardized HGD proportions per lab- oratory. The horizontal line illustrates the mean observed HGD pro- laboratories. Yet, as colorectal polyps are approximately four portion, which is 9.4%. Red bars indicate academic laboratories. Blue times more common than duodenal polyps [23, 24], it can be bars indicate general laboratories. HGD high grade dysplasia. (Color expected that the overall volume of graded polyps will make figure online). a larger difference in our study, especially as the differences between low- and high-volume laboratories are larger. As an explanation, Kuijpers et al. mention that the subjective Fig. 3 Funnel plot representing the variance between all laboratories 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… Table 2 Characteristics of top 3 Laboratories with low- Laboratories with highest standardized HGD proportion and bottom 3 laboratories est standardized HGD 321 reports (n = 3) proportion 212 reports (n = 3) Type of laboratory, n (%)  Academic 0 (0) 2 (66.6)  General 3 (100) 1 (33.3)  Number of patients, 35.7 (25–42) 38.0 (24–63) mean (min–max)  Number of reports, 70.7 (55–91) 101.7 (55—165) mean (min–max) Previous assessment 201 (94.8) 299 (93.1) by different labora- tory, n (%)  No Degree of dysplasia, n (%)  No dysplasia 5 (2.4) 18 (5.6)  LGD 202 (95.3) 256 (79.8)  HGD 4 (1.9) 45 (14.0)  CA 1 (0.5) 2 (0.6) Min minimum, max maximum, LGD low grade dysplasia, HGD high grade dysplasia, CA cancer Indicating if a different laboratory previously assessed tissue of the same patient criteria for defining dysplasia leave room for variation in laboratories (55 years, p = 0.04). Since patients in high-vol- interpretation among diagnosing pathologists, which will ume (referral) centers are older, it seems logical that their likely also increase laboratory variation. duodenal disease was more extensive as it is known that the severity of duodenal polyposis increases with age [26]. Explaining the variation In addition, high-volume laboratories reported more speci- mens per report than low-volume laboratories (1.31 vs 1.15, In a first attempt to explain the variation we corrected for p = 0.039), probably as a result of more extensive duodenal case-mix, which reduced the variation by approximately disease. This may lead to a higher probability of diagnosing 50%, as shown by a reduction of the factor score from 7.8 to HGD. Second, it may also be possible that a pathologist 3.9. This indicates that characteristics of the patient popu- working in a small volume laboratory and therefore less fre- lations varied between laboratories. Our data also showed quently examining duodenal adenomas has more difficulties that more reports were included from the three highest HGD with diagnosing HGD; however solid evidence for this is as diagnosing laboratories compared to the three lowest HGD far as we know not available. diagnosing laboratories (101.7 vs. 70.7 reports, respec- Based on current literature and guidelines, there are tively). In addition, only 34.1% of the reports were included three other possible explanations to explain the variation in from 17 general laboratories, whereas 65.9% of the reports diagnosing HGD in the present study. First, (inter)national were included from 8 academic laboratories. This indicates guidelines vary regarding the procedure and timing to sam- that small volume (mostly general) laboratories diagnosed ple duodenal tissue. For instance, the European Society for HGD less frequently in duodenal tissue from FAP patients Gastrointestinal Endoscopy (ESGE) was the first in 2019 than large volume (mostly academic) laboratories. to discourage routine biopsies of suspected lesions in the Differences in HGD proportions between large- and duodenum for FAP patients, as this may cause fibrosis small volume laboratories might be further explained in which may lead to difficulties in future possible endoscopic two ways. First, international guidelines recommend that resection [9]. But even before 2019 this was already done in FAP patients with extensive duodenal or ampullary disease some hospitals [27]. In addition, conflicting recommenda- should be referred to high-volume expert centers to consider tions exist regarding (endoscopic or surgical) resection of (endoscopic or surgical) resection [9, 10, 25]. Our descrip- duodenal tissue. For example, the ESGE recommends polyp tive data showed that patients in high-volume laboratories size ≥ 10 mm as indication for endoscopic resection, while had an older mean age (60 years) than those in low-volume the Netherlands Foundation for Detection of Hereditary 1 3 E. Soons et al. Tumors recommends resection when Spigelman stage IV be involved in the diagnostic process, which is expected disease, HGD or growing papillary adenomas are found [9, to reduce interobserver variability [29, 38]. Moreover, as 28]. This varying recommendations between guidelines will suggested in previous studies, multidisciplinary team meet- probably lead to variation in tissue sampling between hospi- ings may further reduce interobserver variability [32, 39, tals. In turn, this might have led to differences in the quantity 40]. Future research should show if laboratory variation has of duodenal tissue to be graded and hence the probability of indeed reduced when histopathological diagnosis in mainly diagnosing HGD per laboratory. performed in expert centers. Second, criteria to grade dysplasia are subjective and It is clear that too much subjectivity in diagnosing HGD depending on the interpretation by pathologists. Kuijpers is unwarranted. We therefore encourage better standardiza- et al. performed a questionnaire study that showed consider- tion of histologic grading criteria for duodenal adenomas. In able heterogeneity in the criteria applied by pathologists to addition, previous literature has shown that the implemen- grade dysplasia within colorectal adenomas [19]. As a pos- tation of an e-learning improves interobserver variability sible consequence, several studies evaluating interobserver in Dutch laboratories regarding the grading of colorectal variability in dysplasia grading of colorectal adenomas have dysplasia [41]. There is no reason to believe that widespread shown widely varying results from poor to good agreement implementation of e-learnings also will decrease variation in between pathologists (κ = 0.02–0.69) [29–35]. Subjectivity grading of duodenal dysplasia. Furthermore, the use of arti- within a diagnosing guideline can lead to both under- and ficial intelligence has the potential to decrease variation in overdiagnosis of HGD as pathologists might suffer from HGD diagnosis. However, current research is limited to the ‘professional uncertainty’. The latter is hypothesized to recognition of colorectal dysplasia (without subdividing it occur when physicians are uncertain about a clinical deci- into low- or high grade) and carcinomas [42, 43]. Therefore, sion [36, 37]. For pathologists this means that heterogene- future research is warranted to investigate the role of artifi- ity in diagnostic criteria for HGD may lead to insecurity in cial intelligence in diagnosing HGD in duodenal adenomas. diagnosing it, which in turn may lead to variation in HGD It is also important to make current guidelines on poly- diagnosis. posis syndromes more consistent regarding taking biopsies Third, over the years (inter)national and even local pro- from duodenal polyps, and to define a uniform clinical strat- tocols have been inconsistent regarding the clinical con- egy when HGD in duodenal polyps is diagnosed (i.e., HGD sequences of a HGD diagnosis for duodenal tissue. The as an indication for endoscopic or surgical interventions American Society for Gastrointestinal Endoscopy (ASGE) or not). The clinical guidelines from the European Heredi- guideline recommends considering endoscopic therapy for tary Tumour Group (EHTG) on polyposis syndromes are a lesion with HGD, whereas the ESGE guideline does not currently being revised. This gives the opportunity for at give a clinical recommendation when HGD is diagnosed, least European guidelines to become more uniform in their even though HGD is regarded as a risk factor for devel- recommendations. oping duodenal cancer [9, 11]. Moreover, long-term data on the effect of endoscopic resection of duodenal polyposis Strengths and limitations in FAP patients is only limited available [27]. Though we were unable to collect information on the local protocols for Our study has several strengths and limitations. A definite management of a HGD diagnosis per hospital, it is likely strength is that we were able to use nationwide, longitudinal that these differed between hospitals. It may well be that data, including a cohort of 1217 patients (5782 pathology this variation in local protocols (further) causes professional reports), which is large as FAP is a rare disorder. Second, we uncertainty as pathologists are uncertain about the subse- were able to show laboratory variation in clinical practice, quent consequences after HGD is diagnosed, leading to both rather than in a controlled study design, as was the case in over- and underdiagnosis of HGD. previous interobserver variability studies [29–35]. In addition, some limitations should be addressed as well. Future prospects First, inherent to the data source, specified clinical data on patient characteristics (e.g., age at first upper endoscopy and Various developments have been implemented to improve genetic mutation), endoscopic findings (e.g., size of polyps the diagnosis of HGD in pathology laboratories. Recently, observed during upper endoscopy) and laboratory specifi- five FAP expertise centers in the Netherlands were selected cations (e.g., practices of double reading) were not avail- to clinically manage the disease, including the histological able. These characteristics are all known to be predictive diagnosis of HGD. It can be expected that this will decrease factors for developing HGD and duodenal cancer [13, 14]. the interlaboratory variation and misdiagnoses of HGD and Nevertheless, it is known that endoscopic characteristics of increase uniformity in HGD grading. Due to centralization, duodenal polyposis in FAP patients are poorly reported, as only dedicated gastroenterologists and pathologists will the multiplicity of the polyps impedes exact documentation 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… Informed consent All PALGA data are pseudonymized by a trusted of number, size and location of the polyps. It is unknown third party, securing that in the PALGA database no personally identifi- whether this additional information from endoscopy reports able data are collected. Therefore we were unable to obtain informed would have changed our findings. Second, while we created consent from individual patients. However, data from patients who a large cohort of 1217 patients, the data were collected over refuse their data to be used for scientific research are excluded from the PALGA database. a period of 29 years. However, our sensitivity analysis did not show any remarkable differences, which indicates that Open Access This article is licensed under a Creative Commons Attri- our results are fairly robust. bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes Conclusion were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated Laboratory variation in histological grading of duodenal otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not adenomas of FAP patients was found to be moderate. Patient permitted by statutory regulation or exceeds the permitted use, you will characteristics considerably explained the variation, indi- need to obtain permission directly from the copyright holder. To view a cating patient populations differed between hospitals. Still, copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . there is considerable variation, which leaves room for quality improvement. We are optimistic that the nationwide labora- tory variation will decrease with the centralization of care for References patients with FAP in five expertise centers in the Netherlands. However, further standardization of the grading criteria for 1. Jasperson KW, Tuohy TM, Neklason DW et al (2010) Hereditary and familial colon cancer. Gastroenterology 138:2044–2058 dysplasia of gastro-intestinal and thus duodenal adenomas is 2. Kanth P, Grimmett J, Champine M et al (2017) Hereditary colo- needed and (inter)national guidelines should become more rectal polyposis and cancer syndromes: a primer on diagnosis and uniform regarding the necessity to routinely take biopsies from management. Am J Gastroenterol 112:1509–1525 duodenal polyps and on the clinical consequences of a HGD 3. Bodmer WF, Bailey CJ, Bodmer J et al (1987) Localization of the gene for familial adenomatous polyposis on chromosome 5. diagnosis in FAP patients to decrease unwarranted laboratory Nature 328:614–616 variation. 4. Powell SM, Petersen GM, Krush AJ et  al (1993) Molecular diagnosis of familial adenomatous polyposis. N Engl J Med Supplementary Information The online version contains supplemen- 329:1982–1987 tary material available at https://doi. or g/10. 1007/ s10689- 022- 00320-1 . 5. Karstensen JG, Burisch J, Pommergaard HC et al (2019) Colo- rectal cancer in individuals with familial adenomatous polyposis, Acknowledgements We thank Chantal Kuijpers for her assistance based on analysis of the Danish polyposis registry. Clin Gastro- and expertise to collect our data from PALGA, the Dutch nationwide enterol Hepatol 17:2294-2300.e1 pathology databank. 6. Belchetz LA, Berk T, Bapat BV et al (1996) Changing causes of mortality in patients with familial adenomatous polyposis. Dis Author contributions ES designed and performed the study, analyzed Colon Rectum 39:384–387 and interpreted the data, and wrote the first draft of the paper. FA 7. Offerhaus GJ, Giardiello FM, Krush AJ et al (1992) The risk of supervised the design of the study and the statistical analyses, and upper gastrointestinal cancer in familial adenomatous polyposis. contributed to the interpretation of the results and critically reviewed Gastroenterology 102:1980–1982 the manuscript, CvdP, IN, TB, MvK and PS also contributed to the 8. Syngal S, Brand RE, Church JM, et al (2015) ACG clinical guide- conception and design of the study, interpretation of the results and line: Genetic testing and management of hereditary gastrointesti- critically reviewed the manuscript. LvL collected and analyzed the nal cancer syndromes. Am J Gastroenterol 110:223–262;quiz 263. data. All authors approved the final version to be published. 9. van Leerdam ME, Roos VH, van Hooft JE et al (2019) Endo- scopic management of polyposis syndromes: European Society Funding No funding was received for conducting this study. of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy 51:877–895 Data availability The datasets generated during and/or analysed dur- 10. Monahan KJ, Bradshaw N, Dolwani S et al (2020) Guidelines for ing the current study are available from the corresponding author on the management of hereditary colorectal cancer from the British reasonable request. Society of Gastroenterology (BSG)/Association of Coloproctol- ogy of Great Britain and Ireland (ACPGBI)/United Kingdom Declarations Cancer Genetics Group (UKCGG). Gut 69:411–444 11. Yang J, Gurudu SR, Koptiuch C, et al (2020) American Society for Conflict of interest The authors have no relevant financial or non-fi- Gastrointestinal Endoscopy guideline on the role of endoscopy in nancial interests to disclose. familial adenomatous polyposis syndromes. Gastrointest Endosc 12. Spigelman AD, Williams CB, Talbot IC et al (1989) Upper gastro- Ethical approval Non-identifiable data makes this study to be exempted intestinal cancer in patients with familial adenomatous polyposis. from ethical approval. The scientific and privacy committee of PALGA Lancet 2:783–785 approved the protocol of this study (Reference Number: 2020-41). 13. Thiruvengadam SS, Lopez R, O’Malley M et al (2019) Spigel- man stage IV duodenal polyposis does not precede most duodenal 1 3 E. Soons et al. cancer cases in patients with familial adenomatous polyposis. 31. Foss F, Milkins S, McGregor A (2012) Inter-observer variabil- Gastrointest Endosc 89:345-354.e2 ity in the histological assessment of colorectal polyps detected 14. Sourrouille I, Lefevre JH, Shields C et al (2017) Surveillance of through the NHS Bowel Cancer Screening Programme. Histopa- duodenal polyposis in familial adenomatous polyposis: should the thology 61:47–52 Spigelman score be modified? Dis Colon Rectum 60:1137–1146 32. Turner JK, Williams GT, Morgan M et al (2013) Interobserver 15. Roos VH, Bastiaansen BAJ, Dekker E (2019) Challenges and agreement in the reporting of colorectal polyp pathology among pitfalls of investigating duodenal cancer in patients with familial bowel cancer screening pathologists in Wales. Histopathology adenomatous polyposis. Gastrointest Endosc 89:355–356 62:916–924 16. Westert GP, Groenewoud S, Wennberg JE et al (2018) Medical 33. Mahajan D, Downs-Kelly E, Liu X et al (2013) Reproducibility practice variation: public reporting a first necessary step to spark of the villous component and high-grade dysplasia in colorectal change. Int J Qual Health Care 30:731–735 adenomas <1 cm: implications for endoscopic surveillance. Am 17. Wennberg JE (2002) Unwarranted variations in healthcare deliv- J Surg Pathol 37:427–433 ery: implications for academic medical centres. BMJ 325:961–964 34. Lasisi F, Mouchli A, Riddell R et al (2013) Agreement in inter- 18. Schlemper RJ, Riddell RH, Kato Y et al (2000) The Vienna clas- preting villous elements and dysplasia in adenomas less than one sification of gastrointestinal epithelial neoplasia. Gut 47:251–255 centimetre in size. Dig Liver Dis 45:1049–1055 19. Kuijpers CC, Sluijter CE, von der Thusen JH et al (2016) Inter- 35. Osmond A, Li-Chang H, Kirsch R et al (2014) Interobserver vari- laboratory variability in the grading of dysplasia in a nationwide ability in assessing dysplasia and architecture in colorectal adeno- cohort of colorectal adenomas. Histopathology 69:187–197 mas: a multicentre Canadian study. J Clin Pathol 67:781–786 20. Casparie M, Tiebosch AT, Burger G et al (2007) Pathology data- 36. Birkmeyer JD, Reames BN, McCulloch P et al (2013) Under- banking and biobanking in The Netherlands, a central role for standing of regional variation in the use of surgery. Lancet PALGA, the nationwide histopathology and cytopathology data 382:1121–1129 network and archive. Cell Oncol 29:19–24 37. Wennberg J, Gittelsohn (1973) Small area variations in health care 21. Wennberg JE (2011) Time to tackle unwarranted variations in delivery. Science 182:1102–1108. practice. BMJ 342:d1513 38. Denis B, Peters C, Chapelain C et al (2009) Diagnostic accuracy 22. Roos VH, Bastiaansen BA, Kallenberg FGJ, et al (2020) Endo- of community pathologists in the interpretation of colorectal pol- scopic management of duodenal adenomas in patients with famil- yps. Eur J Gastroenterol Hepatol 21:1153–1160 ial adenomatous polyposis. Gastrointest Endosc 39. Smits LJH, Vink-Börger E, van Lijnschoten G, et al (2021) Diag- 23. Jepsen JM, Persson M, Jakobsen NO et al (1994) Prospective nostic variability in the histopathological assessment of advanced study of prevalence and endoscopic and histopathologic charac- colorectal adenomas and early colorectal cancer in a screening teristics of duodenal polyps in patients submitted to upper endos- population. Histopathology copy. Scand J Gastroenterol 29:483–487 40. Rampioni Vinciguerra GL, Antonelli G, Citron F et al (2019) 24. Pan J, Cen L, Xu L et al (2020) Prevalence and risk factors for Pathologist second opinion significantly alters clinical manage- colorectal polyps in a Chinese population: a retrospective study. ment of pT1 endoscopically resected colorectal cancer. Virchows Sci Rep 10:6974 Arch 475:665–668 25. Yang JG, Koptiuch C, Agrawal D, Buxbaum JL, Abbas Fehmi 41. Madani A, Kuijpers C, Sluijter CE et al (2019) Decrease of vari- SM, Fishman DS, Khashab MA, Jamil LH, Jue TL, Law JK, Lee ation in the grading of dysplasia in colorectal adenomas with a JK, Naveed M, Qumseya BJ, Sawhney MS, Thosani N, Wani national e-learning module. Histopathology 74:925–932 SB, Samadder NJ (2020) American Society for Gastrointestinal 42. Wang KS, Yu G, Xu C et al (2021) Accurate diagnosis of colorec- Endoscopy guideline on the role of endoscopy in familial adeno- tal cancer based on histopathology images using artificial intel- matous polyposis syndromes. Gastrointest Endosc ligence. BMC Med 19:76 26. Bulow S, Bjork J, Christensen IJ et al (2004) Duodenal adenoma- 43. Ho C, Zhao Z, Chen XF et al (2022) A promising deep learning- tosis in familial adenomatous polyposis. Gut 53:381–386 assistive algorithm for histopathological screening of colorectal 27. Soons E, Bisseling TM, van Kouwen MCA et al (2021) Endo- cancer. Sci Rep 12:2222 scopic management of duodenal adenomatosis in familial adeno- matous polyposis—a case-based review. United Eur Gastroenterol Publisher's Note Springer Nature remains neutral with regard to J 9:461–468 jurisdictional claims in published maps and institutional affiliations. 28. Tumors TNFfDoH (2017) Erfelijke Familiaire Tumoren - Richtli- jnen voor diagnostiek en preventie. 29. Terry MB, Neugut AI, Bostick RM et al (2002) Reliability in the classification of advanced colorectal adenomas. Cancer Epidemiol Biomark Prevent 11:660–663 30. van Putten PG, Hol L, van Dekken H et al (2011) Inter-observer variation in the histological diagnosis of polyps in colorectal can- cer screening. Histopathology 58:974–981 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Familial Cancer Springer Journals

Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous polyposis patients

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/laboratory-variation-in-the-grading-of-dysplasia-of-duodenal-adenomas-qAjYKShUZT

References (51)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
ISSN
1389-9600
eISSN
1573-7292
DOI
10.1007/s10689-022-00320-1
Publisher site
See Article on Publisher Site

Abstract

To prevent duodenal and ampullary cancer in familial adenomatous polyposis (FAP) patients, a diagnosis of high grade dysplasia (HGD) plays an important role in the clinical management. Previous research showed that FAP patients are both over- and undertreated after a misdiagnosis of HGD, indicating unwarranted variation. We aimed to investigate the labora- tory variation in dysplasia grading of duodenal adenomas and explore possible explanations for this variation. We included data from all Dutch pathology laboratories between 1991 and 2020 by retrieving histology reports from upper endoscopy specimens of FAP patients from the Dutch nationwide pathology databank (PALGA). Laboratory variation was investigated by comparing standardized proportions of HGD. To describe the degree of variation between the laboratories a factor score was calculated. A funnel plot was used to identify outliers. A total of 3050 specimens from 25 laboratories were included in the final analyses. The mean observed HGD proportion was 9.4%. The top three HGD-diagnosing laboratories diagnosed HGD 3.9 times more often than the lowest three laboratories, even after correcting for case-mix. No outliers were identified. Moderate laboratory variation was found in HGD diagnoses of duodenal tissue of FAP patients after adjusting for case-mix. Despite the fact that no outliers were observed, there may well be room for quality improvement. Concentration of these patients in expertise centers may decrease variation. To further reduce unwarranted variation, we recommend (inter)national guidelines to become more uniform in their recommendations regarding duodenal tissue sampling and consequences of HGD diagnoses. Keywords Pathology · Gastroenterology · Familial adenomatous polyposis (FAP) · Duodenal polyposis Introduction formation of ≥ 100 synchronous polyps distributed through- out the gastrointestinal tract [3, 4]. The standardization of Familial adenomatous polyposis (FAP) affects one in 10,000 colonic screening at a young age and subsequent preven- individuals, which makes it the second most common inher- tive surgery has largely reduced mortality from CRC [5]. ited colorectal cancer (CRC) syndrome [1, 2]. A mutation In addition, individuals with FAP have an increased risk of in the adenomatous polyposis coli (APC) gene leads to the duodenal and ampullary cancer, with relative risks of 331 and 124, respectively, compared to the general population. Consequently, duodenal and ampullary cancer are nowadays * E. Soons the most common cause of cancer-related death in FAP [6, elsa.soons@radboudumc.nl 7]. The histological grade of dysplasia in duodenal adeno- Department of Gastroenterology and Hepatology, Radboud mas plays an important role in the decision-making process Institute for Health Sciences, Radboud University Medical Center, Nijmegen, The Netherlands to prevent duodenal and/or ampullary cancer in two ways. First, international guidelines recommend starting duodenal Department of Pathology, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, surveillance at the age of 25–35 years [8–11]. The surveil- Nijmegen, The Netherlands lance interval is traditionally determined by the Spigelman Department of IQ Healthcare, Radboud Institute for Health classification, which includes histological grading as one Sciences, Radboud University Medical Center, Nijmegen, of four decisive parameters [12]. A diagnosis of high grade The Netherlands Vol.:(0123456789) 1 3 E. Soons et al. dysplasia (HGD) will lead to a shortened surveillance inter- pathology reports from Dutch pathology laboratories, with val in many cases. Secondly, the presence of HGD is a rela- nationwide coverage since 1991 [20]. All PALGA data tive indication for an endoscopic or surgical intervention as are pseudonymized by a trusted third party, securing that it is considered a risk factor for developing duodenal cancer in the PALGA database no personally identifiable data are [9, 11, 13–15]. collected. Data from patients who refuse their data to be As a result of the need to screen for dysplasia in duodenal used for scientific research are excluded from the PALGA tissue in FAP patients, duodenal tissue is routinely seen by database. The scientific and privacy committee of PALGA pathologists. Yet, there are indications that misdiagnoses approved the protocol of this study (Reference Number: of dysplasia occur, which may have major clinical conse- 2020-41). Non-identifiable data makes this study to be quences. For instance, Sourouille et al. described the histo- exempted from ethical approval. pathological diagnoses of 52 duodenal specimens collected We identified all reports with one or more diagnoses of after radical surgical treatment [14]. Surgery was performed duodenal adenoma between 1991 and 2020 from patients in FAP patients with endoscopically untreatable duodenal diagnosed with FAP or who had a prior (sub)total colectomy, polyposis and/or an ampullary adenoma or Spigelman score assuming they were also FAP-patients. If there were multiple IV with HGD at 2 successive endoscopic assessments three records per patients, all reports were included. months apart and confirmed by two independent patholo- gists. They found that in eleven patients (21.2%) surgery was Laboratory, patient and specimen selection performed too late (i.e., cancer was already present), while in fourteen patients (29.7%) surgery was performed too early Based on the search criteria described above, a total of (i.e., no HGD or cancer was present). Both scenarios indi- 5782 reports from 1217 patients and 49 laboratories were cate inappropriate care, as the misdiagnosis of HGD led to identified. To account for small sample variations, reports both over- and undertreatment of duodenal polyposis. from laboratories with < 30 reports (n = 266) or without an Measuring institutional variation in clinical care may help HGD diagnosis (n = 33) over the total inclusion period were identifying inappropriate care and/or suboptimal quality of excluded from further analysis (see Fig. 1). To aim for uni- care and may provide target points for quality improvement formity in our dataset, pathology reports were also excluded [16]. Variation in clinical care is an issue, especially if it if they were inconclusive on the degree of dysplasia (n = 109) is unwarranted. The latter may occur in clinical care when patients undergo care which is not indicated, as illustrated in the above-mentioned study by Sourouille et al. [17]. So far, no studies on laboratory variation in the grading of dysplasia in duodenal tissue have been published. In clinical practice, the guideline that is used for grading duodenal dysplasia is the same as for colorectal dysplasia [18]. In the latter, considerable interlaboratory variation has been reported in grading of colonic dysplasia, with 35% of laboratories reporting a significantly lower or higher frequency of HGD than average; however this information is not yet known for duodenal dysplasia grading [19]. If the same is true for duo- denal tissue, there is need to reduce this variation to prevent over- and undertreatment of duodenal polyposis. The aim of the current study was to investigate the extent of laboratory variation in the histological dysplasia grading of duodenal adenomas from patients with FAP in a nation- wide cohort and, if present, to identify possible explanations for this variation. Methods Data extraction All data were extracted from PALGA, the Dutch nation- Fig. 1 Flowchart representing reasons for exclusion for excluded wide pathology databank. PALGA contains excerpts of all reports 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… or origin of the tissue (n = 2101), revised cases (n = 7), from Laboratory variation in proportion HGD diagnosis patients < 18 years of age (n = 11) or from patients who pre- viously underwent duodenal resection (n = 137). Further- Laboratories were compared by indirect standardization of more, reports of resection specimens were excluded (n = 68) HGD proportions. First, the observed proportion of HGD when information on the number, size and location of the diagnoses per laboratory was assessed. Second, expected duodenal specimens remained unclear in these cases. To proportions of HGD diagnoses per laboratory were calcu- correct for multiple paired measurements, we included one lated with a multilevel logistic regression model. This model specimen per report. This was either the (first) specimen accounts for age, year of the report, number of specimens diagnosed with HGD, or, in absence of a HGD diagnosis, per report, localization in the duodenum, localization at the first specimen that was described in the report. the papilla major, morphology and method used to obtain The final dataset used for further analyses therefore the tissue (i.e., polypectomy or other). Localization in the consisted of 3050 specimens from 926 patients in 25 duodenum at the papilla major, or the method of obtaining laboratories. the tissue were significant predictors in univariate analysis. Nevertheless, as previous studies showed the significance of Data collection these variables, they were included in the final model regard- less [14, 19]. See Supplementary Table 1 for the contribu- For each laboratory, we registered the type of laboratory tion of each variable to the final model. Third, the observed (academic or general), number of patients and number of HGD proportions were divided by the expected (adjusted) specimens. For each patient we extracted age, sex and total HGD proportions per laboratory. This led to an observed/ number of specimens. Only the first report per patient was expected-ratio (O/E-ratio) indicating less HGD diagno- used to describe these characteristics since multiple reports ses than expected when < 1, or more HGD diagnoses than per patient were included for most patients (55.5%). For each expected when > 1. Fourth, by multiplying the O/E-ratio specimen extracted, the year of the histology report, total with the overall HGD proportion, the standardized HGD number of removed specimens per report, degree of dyspla- proportion was calculated. sia (no dysplasia, LGD, HGD or carcinoma), morphological To quantify the amount of interlaboratory variation in type (tubular, tubulovillous or villous), localization in the HGD diagnoses, a factor score was calculated. Factor scores duodenum from first to fourth part (D1-D4), tissue from the are used to illustrate by which factor the highest score differs duodenal papilla or not and tissue obtained by biopsy or not, from the lowest score. The factor score was calculated by was recorded. dividing the mean proportion of the highest three labora- Clinical characteristics were extracted manually. To tories by the mean proportion of the lowest 3 laboratories. validate the extracted data, 10% of the reports were double A variation of factor two is considered to be modest [21]. checked by three additional investigators (ES, MvK and TB). Multiple imputation was performed for the variable mor- Funnel plots phology, which had 26.3% missing data. All other variables were complete. Funnel plots were constructed to detect outliers. In short, these are frequently used control charts in which an out- Statistical analysis come measure for a unit of analysis (e.g., a histopathologic laboratory) is plotted against a measure for the laboratory Descriptive analysis size (“precision”). The O/E-ratios were plotted against their expected values and control limits (95% and 99.8%) were Overall laboratory characteristics were described with included around the target value (O/E = 1). O/E-ratios of respect to laboratory type and number of patients and speci- laboratories outside the control limits are considered outli- mens. Overall patient characteristics were described with ers and perform significantly different from the target value. respect to age, sex and number of specimens. Specimen Additionally, laboratories that lie between both control limits characteristics were described with respect to year of his- can be considered as random variation. tology report, total number of removed specimens per report, location in duodenum, degree of dysplasia, morphology type Explaining and understanding laboratory variation and method used to obtain the tissue. Categorical variables in proportion HGD diagnosis are presented as frequencies and percentages. Continuous variables are reported as means ± standard deviation (SD) To study whether case-mix at least partly explained the or medians (interquartile range (IQR)), in case of a skewed laboratory variation, we compared factor scores based on distribution. the standardized proportions with the factors scores of 1 3 E. Soons et al. unstandardized proportions. Any explanatory influence of observed HGD proportion was 9.4%. The highest stand- our case-mix variables should result in lower factor scores. ardized HGD proportion was 14.9%, whereas the lowest In addition, to further explain the variation, laboratories was 3.5%. All academic laboratories reported more HGD with low and high standardized HGD proportions were diagnoses than average. The mean highest 3/lowest 3 factor compared for laboratory type, number of reports, previous score for the standardized HGD proportions was 3.9, which assessment by another laboratory (i.e., if a different labo- indicates that tissue diagnosed in the highest 3 diagnosing ratory previously assessed tissue of the same patient) and laboratories had a 3.9 times higher likelihood of being diag- degree of dysplasia. For this, the three laboratories with the nosed as HGD than tissue diagnosed in a laboratory from the lowest three standardized HGD proportions and the three lowest 3 laboratories. laboratories with highest three standardized HGD propor- Figure  3 presents a funnel plot showing the variation tions were selected. between laboratories. The O/E (i.e., standardized) ratio is presented on the y-axis, and “expected”, the number of Sensitivity analysis expected HGD-cases per laboratory, on the x-axis. The O/E- ratios varied from 0.4 to 1.6. One laboratory (i.e., 4% of all When the Spigelman classification was introduced in 1989, laboratories) were located outside the 95% control limits. the dysplasia grading was originally graded as mild, moder- Nonetheless, all laboratories fell within the 99.8% control ate or severe, which was changed into a two-tiered system limits, according to what was expected. (low-grade dysplasia (LGD) and HGD) to decrease interob- server variability [18]. To study the effect of this change on Explaining and understanding laboratory variation our data, we performed a sensitivity analysis in which only in proportion of duodenal HGD diagnoses data between 2000 and 2020 were included, after the intro- duction of the two-tiered system. Laboratory variation and To investigate the effect of case-mix adjustment on the extent magnitude of variation were calculated as described above. of variation, we compared factor scores based on standard- Analyses were performed with R version 1.3.1073 (R ized proportions with the factors scores of unstandardized Foundation for Statistical Computing, Vienna, Austria. proportions. The mean factor score for the unstandardized URL: http:// www.R- proje ct. org/) and IBM SPSS Statistics HGD proportions was 7.8, which decreased to 3.9 after version 25 (SPSS Inc., Chicago, IL, USA). case-mix adjustment. The lower factor scores for standard- ized proportions indicate that our case-mix only partially (3.9/7.8 = 50%) could explain the observed (unstandardized) Results laboratory variation. To further identify explanations for variation, Table  2 Lab, patient and specimen characteristics shows direct comparisons between laboratories with low standardized HGD proportions (lowest 3 laboratories) and Table 1 shows the laboratory, patient and specimen char- laboratories with high standardized HGD proportions (high- acteristics. Eight of 25 (32.0%) included laboratories were est 3 laboratories). The lowest 3 laboratories were all general academic. Mean number of patients and specimens per laboratories, while the two of the three highest laborato- laboratory was 42 (range 11–127) and 122 (range 27–748), ries were academic laboratories. In the lowest 3 laborato- respectively. Mean age of patients at the time of their first ries a mean of 70.7 reports (range 55–91) were included, report was 55.7 (± 18.3) years, and 43.3% of patients was whereas in the highest 3 laboratories a mean of 101.7 female. A median of 3 (IQR 5) specimens per patient was (range 55–165) reports were included. HGD was diagnosed included in the analyses. Most specimens (95.5%) were his- approximately seven times more frequently in the highest 3 tologically reported from 2001 to 2020. In more than half of laboratories compared to the lowest 3 laboratories (14.0% reports (53.7%), one specimen was described. Most speci- vs 1.9%, respectively). Both highest and lowest 3 laborato- mens (84.2%) came from the descending duodenum (D2), ries particularly diagnosed tissue from patients who had not with a minority (14.4%) located at the papilla major. HGD yet been assessed by another laboratory (93.1% vs 94.8%, was diagnosed in 9.4% of the specimens. Morphology was respectively). described as tubular in 63.7%. Only 15.6% of specimens were obtained by polypectomy. Sensitivity analysis Laboratory variation in proportion of HGD diagnosis Sensitivity analysis did not reveal significant differences Figure  2 shows the standardized HGD proportions per in highest 3/lowest 3 factor scores (4.25 for standardized laboratory as well as the overall mean. The overall mean proportions), which indicates that the change in dysplasia 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… Table 1 Laboratory, patient and Laboratory characteristics N = 25 specimen characteristics Academic laboratory, n (%) 8 (32.0) Patients, mean (min–max) 42 (11–217) Specimens, mean (min–max) 122 (27–748) Patient characteristics N = 926 Age (y), mean ± SD 55.7 (18.3) Sex (female), n (%) 402 (43.3) Number of specimens, median (IQR) 3 (5) Specimen characteristics N = 3050 Year of histology report, n (%)  1991–2000 137 (4.5)  2001–2020 2913 (95.5) Removed specimens per report, n (%)  1 1639 (53.7)  2 698 (22.9)  3 314 (10.3)  4 158 (5.2)  ≥ 5 241 (7.9) Location, n (%)  D1 232 (7.6)  D2 2569 (84.2)  D3/4 249 (8.2) Localization at papilla, n (%) 440 (14.4) Degree of dysplasia, n (%) No dysplasia 172 (5.6)  LGD 2580 (84.6)  HGD 287 (9.4)  CA 11 (0.4) Morphology type, n (%)  Tubular 1431 (63.7)  Tubulo-villous 717 (31.9)  Villous 100 (4.4)  Removed by polypectomy, n (%) 477 (15.6) N number; min minimum; max maximum; y years; SD standard deviation; IQR interquartile range; LGD low grade dysplasia; HGD high grade dysplasia; D1 duodenal bulb, D2 descending duodenum, D3 inferior duodenum, D4 ascending duodenum; for subsequent analysis only one specimen per report was selected grading in 2000 did not significantly influence our data. See HGD diagnosing laboratories also had higher volumes com- Supplementary Table 2 for all factor scores. pared to the lowest three HGD diagnosing laboratories. Perspective Discussion Our results showed an observed HGD-proportion of 9.4%. This study in a Dutch nationwide cohort shows that there In addition, HGD was diagnosed in 7.0% and 10.7% reports is moderate laboratory variation in scoring HGD in duode- from general versus academic laboratories, respectively. Pre- nal adenomas of FAP patients, as indicated by standardized vious literature on the prevalence of HGD in duodenal pol- HGD proportions. Additional explanatory analyses showed yps in FAP patients is scarce. However, two recent studies that the case-mix of this study could explain approximately briefly discussed this. First, Sourrouille et al. reported that half of the observed laboratory variation. The highest three 3.9% of their included cases were diagnosed with HGD after the first upper endoscopy. Five and 10-year rates of HGD 1 3 E. Soons et al. were 12.1% and 20.8%, respectively [14]. Second, Roos et al. reported that 4% and 17% of endoscopically removed polyps from the duodenum and papilla were diagnosed as HGD, respectively [22]. Both studies were performed in tertiary centers. Therefore, our results (both overall HGD-proportion as well as the HGD-proportion in academic laboratories) fell within the previously reported range of HGD prevalences. To our knowledge, this is the first study to report the labo- ratory variation in HGD scoring of duodenal adenomas in patients with FAP. It is important to realize that the same guideline is currently used for the grading of dysplasia for both colorectal and duodenal tissue [18]. A Dutch study by Kuijpers et al. showed considerable laboratory variability in dysplasia grading of colorectal adenomas, as illustrated by the fact that 13 of 37 (35%) included laboratories were aber- rant, i.e. they reported a significantly lower or higher fre- quency of HGD in colorectal adenomas than average, even after correcting for case-mix. Most of these aberrant labo- ratories (9/13, 69.2%) reported more HGD than expected, which is in line with our results as standardized proportions of HGD were higher than average in 19/25 (76.0%) labora- tories. The effect of volume differences per laboratory was not analyzed in this study [19]. Our results show that the highest 3 laboratories graded more polyps than the lowest 3 Fig. 2 Bar chart representing standardized HGD proportions per lab- oratory. The horizontal line illustrates the mean observed HGD pro- laboratories. Yet, as colorectal polyps are approximately four portion, which is 9.4%. Red bars indicate academic laboratories. Blue times more common than duodenal polyps [23, 24], it can be bars indicate general laboratories. HGD high grade dysplasia. (Color expected that the overall volume of graded polyps will make figure online). a larger difference in our study, especially as the differences between low- and high-volume laboratories are larger. As an explanation, Kuijpers et al. mention that the subjective Fig. 3 Funnel plot representing the variance between all laboratories 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… Table 2 Characteristics of top 3 Laboratories with low- Laboratories with highest standardized HGD proportion and bottom 3 laboratories est standardized HGD 321 reports (n = 3) proportion 212 reports (n = 3) Type of laboratory, n (%)  Academic 0 (0) 2 (66.6)  General 3 (100) 1 (33.3)  Number of patients, 35.7 (25–42) 38.0 (24–63) mean (min–max)  Number of reports, 70.7 (55–91) 101.7 (55—165) mean (min–max) Previous assessment 201 (94.8) 299 (93.1) by different labora- tory, n (%)  No Degree of dysplasia, n (%)  No dysplasia 5 (2.4) 18 (5.6)  LGD 202 (95.3) 256 (79.8)  HGD 4 (1.9) 45 (14.0)  CA 1 (0.5) 2 (0.6) Min minimum, max maximum, LGD low grade dysplasia, HGD high grade dysplasia, CA cancer Indicating if a different laboratory previously assessed tissue of the same patient criteria for defining dysplasia leave room for variation in laboratories (55 years, p = 0.04). Since patients in high-vol- interpretation among diagnosing pathologists, which will ume (referral) centers are older, it seems logical that their likely also increase laboratory variation. duodenal disease was more extensive as it is known that the severity of duodenal polyposis increases with age [26]. Explaining the variation In addition, high-volume laboratories reported more speci- mens per report than low-volume laboratories (1.31 vs 1.15, In a first attempt to explain the variation we corrected for p = 0.039), probably as a result of more extensive duodenal case-mix, which reduced the variation by approximately disease. This may lead to a higher probability of diagnosing 50%, as shown by a reduction of the factor score from 7.8 to HGD. Second, it may also be possible that a pathologist 3.9. This indicates that characteristics of the patient popu- working in a small volume laboratory and therefore less fre- lations varied between laboratories. Our data also showed quently examining duodenal adenomas has more difficulties that more reports were included from the three highest HGD with diagnosing HGD; however solid evidence for this is as diagnosing laboratories compared to the three lowest HGD far as we know not available. diagnosing laboratories (101.7 vs. 70.7 reports, respec- Based on current literature and guidelines, there are tively). In addition, only 34.1% of the reports were included three other possible explanations to explain the variation in from 17 general laboratories, whereas 65.9% of the reports diagnosing HGD in the present study. First, (inter)national were included from 8 academic laboratories. This indicates guidelines vary regarding the procedure and timing to sam- that small volume (mostly general) laboratories diagnosed ple duodenal tissue. For instance, the European Society for HGD less frequently in duodenal tissue from FAP patients Gastrointestinal Endoscopy (ESGE) was the first in 2019 than large volume (mostly academic) laboratories. to discourage routine biopsies of suspected lesions in the Differences in HGD proportions between large- and duodenum for FAP patients, as this may cause fibrosis small volume laboratories might be further explained in which may lead to difficulties in future possible endoscopic two ways. First, international guidelines recommend that resection [9]. But even before 2019 this was already done in FAP patients with extensive duodenal or ampullary disease some hospitals [27]. In addition, conflicting recommenda- should be referred to high-volume expert centers to consider tions exist regarding (endoscopic or surgical) resection of (endoscopic or surgical) resection [9, 10, 25]. Our descrip- duodenal tissue. For example, the ESGE recommends polyp tive data showed that patients in high-volume laboratories size ≥ 10 mm as indication for endoscopic resection, while had an older mean age (60 years) than those in low-volume the Netherlands Foundation for Detection of Hereditary 1 3 E. Soons et al. Tumors recommends resection when Spigelman stage IV be involved in the diagnostic process, which is expected disease, HGD or growing papillary adenomas are found [9, to reduce interobserver variability [29, 38]. Moreover, as 28]. This varying recommendations between guidelines will suggested in previous studies, multidisciplinary team meet- probably lead to variation in tissue sampling between hospi- ings may further reduce interobserver variability [32, 39, tals. In turn, this might have led to differences in the quantity 40]. Future research should show if laboratory variation has of duodenal tissue to be graded and hence the probability of indeed reduced when histopathological diagnosis in mainly diagnosing HGD per laboratory. performed in expert centers. Second, criteria to grade dysplasia are subjective and It is clear that too much subjectivity in diagnosing HGD depending on the interpretation by pathologists. Kuijpers is unwarranted. We therefore encourage better standardiza- et al. performed a questionnaire study that showed consider- tion of histologic grading criteria for duodenal adenomas. In able heterogeneity in the criteria applied by pathologists to addition, previous literature has shown that the implemen- grade dysplasia within colorectal adenomas [19]. As a pos- tation of an e-learning improves interobserver variability sible consequence, several studies evaluating interobserver in Dutch laboratories regarding the grading of colorectal variability in dysplasia grading of colorectal adenomas have dysplasia [41]. There is no reason to believe that widespread shown widely varying results from poor to good agreement implementation of e-learnings also will decrease variation in between pathologists (κ = 0.02–0.69) [29–35]. Subjectivity grading of duodenal dysplasia. Furthermore, the use of arti- within a diagnosing guideline can lead to both under- and ficial intelligence has the potential to decrease variation in overdiagnosis of HGD as pathologists might suffer from HGD diagnosis. However, current research is limited to the ‘professional uncertainty’. The latter is hypothesized to recognition of colorectal dysplasia (without subdividing it occur when physicians are uncertain about a clinical deci- into low- or high grade) and carcinomas [42, 43]. Therefore, sion [36, 37]. For pathologists this means that heterogene- future research is warranted to investigate the role of artifi- ity in diagnostic criteria for HGD may lead to insecurity in cial intelligence in diagnosing HGD in duodenal adenomas. diagnosing it, which in turn may lead to variation in HGD It is also important to make current guidelines on poly- diagnosis. posis syndromes more consistent regarding taking biopsies Third, over the years (inter)national and even local pro- from duodenal polyps, and to define a uniform clinical strat- tocols have been inconsistent regarding the clinical con- egy when HGD in duodenal polyps is diagnosed (i.e., HGD sequences of a HGD diagnosis for duodenal tissue. The as an indication for endoscopic or surgical interventions American Society for Gastrointestinal Endoscopy (ASGE) or not). The clinical guidelines from the European Heredi- guideline recommends considering endoscopic therapy for tary Tumour Group (EHTG) on polyposis syndromes are a lesion with HGD, whereas the ESGE guideline does not currently being revised. This gives the opportunity for at give a clinical recommendation when HGD is diagnosed, least European guidelines to become more uniform in their even though HGD is regarded as a risk factor for devel- recommendations. oping duodenal cancer [9, 11]. Moreover, long-term data on the effect of endoscopic resection of duodenal polyposis Strengths and limitations in FAP patients is only limited available [27]. Though we were unable to collect information on the local protocols for Our study has several strengths and limitations. A definite management of a HGD diagnosis per hospital, it is likely strength is that we were able to use nationwide, longitudinal that these differed between hospitals. It may well be that data, including a cohort of 1217 patients (5782 pathology this variation in local protocols (further) causes professional reports), which is large as FAP is a rare disorder. Second, we uncertainty as pathologists are uncertain about the subse- were able to show laboratory variation in clinical practice, quent consequences after HGD is diagnosed, leading to both rather than in a controlled study design, as was the case in over- and underdiagnosis of HGD. previous interobserver variability studies [29–35]. In addition, some limitations should be addressed as well. Future prospects First, inherent to the data source, specified clinical data on patient characteristics (e.g., age at first upper endoscopy and Various developments have been implemented to improve genetic mutation), endoscopic findings (e.g., size of polyps the diagnosis of HGD in pathology laboratories. Recently, observed during upper endoscopy) and laboratory specifi- five FAP expertise centers in the Netherlands were selected cations (e.g., practices of double reading) were not avail- to clinically manage the disease, including the histological able. These characteristics are all known to be predictive diagnosis of HGD. It can be expected that this will decrease factors for developing HGD and duodenal cancer [13, 14]. the interlaboratory variation and misdiagnoses of HGD and Nevertheless, it is known that endoscopic characteristics of increase uniformity in HGD grading. Due to centralization, duodenal polyposis in FAP patients are poorly reported, as only dedicated gastroenterologists and pathologists will the multiplicity of the polyps impedes exact documentation 1 3 Laboratory variation in the grading of dysplasia of duodenal adenomas in familial adenomatous… Informed consent All PALGA data are pseudonymized by a trusted of number, size and location of the polyps. It is unknown third party, securing that in the PALGA database no personally identifi- whether this additional information from endoscopy reports able data are collected. Therefore we were unable to obtain informed would have changed our findings. Second, while we created consent from individual patients. However, data from patients who a large cohort of 1217 patients, the data were collected over refuse their data to be used for scientific research are excluded from the PALGA database. a period of 29 years. However, our sensitivity analysis did not show any remarkable differences, which indicates that Open Access This article is licensed under a Creative Commons Attri- our results are fairly robust. bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes Conclusion were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated Laboratory variation in histological grading of duodenal otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not adenomas of FAP patients was found to be moderate. Patient permitted by statutory regulation or exceeds the permitted use, you will characteristics considerably explained the variation, indi- need to obtain permission directly from the copyright holder. To view a cating patient populations differed between hospitals. Still, copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . there is considerable variation, which leaves room for quality improvement. We are optimistic that the nationwide labora- tory variation will decrease with the centralization of care for References patients with FAP in five expertise centers in the Netherlands. However, further standardization of the grading criteria for 1. Jasperson KW, Tuohy TM, Neklason DW et al (2010) Hereditary and familial colon cancer. Gastroenterology 138:2044–2058 dysplasia of gastro-intestinal and thus duodenal adenomas is 2. Kanth P, Grimmett J, Champine M et al (2017) Hereditary colo- needed and (inter)national guidelines should become more rectal polyposis and cancer syndromes: a primer on diagnosis and uniform regarding the necessity to routinely take biopsies from management. Am J Gastroenterol 112:1509–1525 duodenal polyps and on the clinical consequences of a HGD 3. Bodmer WF, Bailey CJ, Bodmer J et al (1987) Localization of the gene for familial adenomatous polyposis on chromosome 5. diagnosis in FAP patients to decrease unwarranted laboratory Nature 328:614–616 variation. 4. Powell SM, Petersen GM, Krush AJ et  al (1993) Molecular diagnosis of familial adenomatous polyposis. N Engl J Med Supplementary Information The online version contains supplemen- 329:1982–1987 tary material available at https://doi. or g/10. 1007/ s10689- 022- 00320-1 . 5. Karstensen JG, Burisch J, Pommergaard HC et al (2019) Colo- rectal cancer in individuals with familial adenomatous polyposis, Acknowledgements We thank Chantal Kuijpers for her assistance based on analysis of the Danish polyposis registry. Clin Gastro- and expertise to collect our data from PALGA, the Dutch nationwide enterol Hepatol 17:2294-2300.e1 pathology databank. 6. Belchetz LA, Berk T, Bapat BV et al (1996) Changing causes of mortality in patients with familial adenomatous polyposis. Dis Author contributions ES designed and performed the study, analyzed Colon Rectum 39:384–387 and interpreted the data, and wrote the first draft of the paper. FA 7. Offerhaus GJ, Giardiello FM, Krush AJ et al (1992) The risk of supervised the design of the study and the statistical analyses, and upper gastrointestinal cancer in familial adenomatous polyposis. contributed to the interpretation of the results and critically reviewed Gastroenterology 102:1980–1982 the manuscript, CvdP, IN, TB, MvK and PS also contributed to the 8. Syngal S, Brand RE, Church JM, et al (2015) ACG clinical guide- conception and design of the study, interpretation of the results and line: Genetic testing and management of hereditary gastrointesti- critically reviewed the manuscript. LvL collected and analyzed the nal cancer syndromes. Am J Gastroenterol 110:223–262;quiz 263. data. All authors approved the final version to be published. 9. van Leerdam ME, Roos VH, van Hooft JE et al (2019) Endo- scopic management of polyposis syndromes: European Society Funding No funding was received for conducting this study. of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy 51:877–895 Data availability The datasets generated during and/or analysed dur- 10. Monahan KJ, Bradshaw N, Dolwani S et al (2020) Guidelines for ing the current study are available from the corresponding author on the management of hereditary colorectal cancer from the British reasonable request. Society of Gastroenterology (BSG)/Association of Coloproctol- ogy of Great Britain and Ireland (ACPGBI)/United Kingdom Declarations Cancer Genetics Group (UKCGG). Gut 69:411–444 11. Yang J, Gurudu SR, Koptiuch C, et al (2020) American Society for Conflict of interest The authors have no relevant financial or non-fi- Gastrointestinal Endoscopy guideline on the role of endoscopy in nancial interests to disclose. familial adenomatous polyposis syndromes. Gastrointest Endosc 12. Spigelman AD, Williams CB, Talbot IC et al (1989) Upper gastro- Ethical approval Non-identifiable data makes this study to be exempted intestinal cancer in patients with familial adenomatous polyposis. from ethical approval. The scientific and privacy committee of PALGA Lancet 2:783–785 approved the protocol of this study (Reference Number: 2020-41). 13. Thiruvengadam SS, Lopez R, O’Malley M et al (2019) Spigel- man stage IV duodenal polyposis does not precede most duodenal 1 3 E. Soons et al. cancer cases in patients with familial adenomatous polyposis. 31. Foss F, Milkins S, McGregor A (2012) Inter-observer variabil- Gastrointest Endosc 89:345-354.e2 ity in the histological assessment of colorectal polyps detected 14. Sourrouille I, Lefevre JH, Shields C et al (2017) Surveillance of through the NHS Bowel Cancer Screening Programme. Histopa- duodenal polyposis in familial adenomatous polyposis: should the thology 61:47–52 Spigelman score be modified? Dis Colon Rectum 60:1137–1146 32. Turner JK, Williams GT, Morgan M et al (2013) Interobserver 15. Roos VH, Bastiaansen BAJ, Dekker E (2019) Challenges and agreement in the reporting of colorectal polyp pathology among pitfalls of investigating duodenal cancer in patients with familial bowel cancer screening pathologists in Wales. Histopathology adenomatous polyposis. Gastrointest Endosc 89:355–356 62:916–924 16. Westert GP, Groenewoud S, Wennberg JE et al (2018) Medical 33. Mahajan D, Downs-Kelly E, Liu X et al (2013) Reproducibility practice variation: public reporting a first necessary step to spark of the villous component and high-grade dysplasia in colorectal change. Int J Qual Health Care 30:731–735 adenomas <1 cm: implications for endoscopic surveillance. Am 17. Wennberg JE (2002) Unwarranted variations in healthcare deliv- J Surg Pathol 37:427–433 ery: implications for academic medical centres. BMJ 325:961–964 34. Lasisi F, Mouchli A, Riddell R et al (2013) Agreement in inter- 18. Schlemper RJ, Riddell RH, Kato Y et al (2000) The Vienna clas- preting villous elements and dysplasia in adenomas less than one sification of gastrointestinal epithelial neoplasia. Gut 47:251–255 centimetre in size. Dig Liver Dis 45:1049–1055 19. Kuijpers CC, Sluijter CE, von der Thusen JH et al (2016) Inter- 35. Osmond A, Li-Chang H, Kirsch R et al (2014) Interobserver vari- laboratory variability in the grading of dysplasia in a nationwide ability in assessing dysplasia and architecture in colorectal adeno- cohort of colorectal adenomas. Histopathology 69:187–197 mas: a multicentre Canadian study. J Clin Pathol 67:781–786 20. Casparie M, Tiebosch AT, Burger G et al (2007) Pathology data- 36. Birkmeyer JD, Reames BN, McCulloch P et al (2013) Under- banking and biobanking in The Netherlands, a central role for standing of regional variation in the use of surgery. Lancet PALGA, the nationwide histopathology and cytopathology data 382:1121–1129 network and archive. Cell Oncol 29:19–24 37. Wennberg J, Gittelsohn (1973) Small area variations in health care 21. Wennberg JE (2011) Time to tackle unwarranted variations in delivery. Science 182:1102–1108. practice. BMJ 342:d1513 38. Denis B, Peters C, Chapelain C et al (2009) Diagnostic accuracy 22. Roos VH, Bastiaansen BA, Kallenberg FGJ, et al (2020) Endo- of community pathologists in the interpretation of colorectal pol- scopic management of duodenal adenomas in patients with famil- yps. Eur J Gastroenterol Hepatol 21:1153–1160 ial adenomatous polyposis. Gastrointest Endosc 39. Smits LJH, Vink-Börger E, van Lijnschoten G, et al (2021) Diag- 23. Jepsen JM, Persson M, Jakobsen NO et al (1994) Prospective nostic variability in the histopathological assessment of advanced study of prevalence and endoscopic and histopathologic charac- colorectal adenomas and early colorectal cancer in a screening teristics of duodenal polyps in patients submitted to upper endos- population. Histopathology copy. Scand J Gastroenterol 29:483–487 40. Rampioni Vinciguerra GL, Antonelli G, Citron F et al (2019) 24. Pan J, Cen L, Xu L et al (2020) Prevalence and risk factors for Pathologist second opinion significantly alters clinical manage- colorectal polyps in a Chinese population: a retrospective study. ment of pT1 endoscopically resected colorectal cancer. Virchows Sci Rep 10:6974 Arch 475:665–668 25. Yang JG, Koptiuch C, Agrawal D, Buxbaum JL, Abbas Fehmi 41. Madani A, Kuijpers C, Sluijter CE et al (2019) Decrease of vari- SM, Fishman DS, Khashab MA, Jamil LH, Jue TL, Law JK, Lee ation in the grading of dysplasia in colorectal adenomas with a JK, Naveed M, Qumseya BJ, Sawhney MS, Thosani N, Wani national e-learning module. Histopathology 74:925–932 SB, Samadder NJ (2020) American Society for Gastrointestinal 42. Wang KS, Yu G, Xu C et al (2021) Accurate diagnosis of colorec- Endoscopy guideline on the role of endoscopy in familial adeno- tal cancer based on histopathology images using artificial intel- matous polyposis syndromes. Gastrointest Endosc ligence. BMC Med 19:76 26. Bulow S, Bjork J, Christensen IJ et al (2004) Duodenal adenoma- 43. Ho C, Zhao Z, Chen XF et al (2022) A promising deep learning- tosis in familial adenomatous polyposis. Gut 53:381–386 assistive algorithm for histopathological screening of colorectal 27. Soons E, Bisseling TM, van Kouwen MCA et al (2021) Endo- cancer. Sci Rep 12:2222 scopic management of duodenal adenomatosis in familial adeno- matous polyposis—a case-based review. United Eur Gastroenterol Publisher's Note Springer Nature remains neutral with regard to J 9:461–468 jurisdictional claims in published maps and institutional affiliations. 28. Tumors TNFfDoH (2017) Erfelijke Familiaire Tumoren - Richtli- jnen voor diagnostiek en preventie. 29. Terry MB, Neugut AI, Bostick RM et al (2002) Reliability in the classification of advanced colorectal adenomas. Cancer Epidemiol Biomark Prevent 11:660–663 30. van Putten PG, Hol L, van Dekken H et al (2011) Inter-observer variation in the histological diagnosis of polyps in colorectal can- cer screening. Histopathology 58:974–981 1 3

Journal

Familial Cancer – Springer Journals

Published: Apr 1, 2023

Keywords: Pathology; Gastroenterology; Familial adenomatous polyposis (FAP); Duodenal polyposis

There are no references for this article.