Incidental findings in the bowel cancer population screening program: other polyps and malignancies - A nationwide study.

IN: designed the study, analysed the data, and wrote the article, MEVB: critically reviewed the article and provided the illustrations, CCHJK: provided and analysed the data and critically reviewed the article, ED and NAS: critically reviewed the article. No funding was obtained for the current study.

From its implementation in 2014, the population between the ages of 55 and 75 biennially receives an invitation and faecal immunochemical test (FIT) for participation in the Dutch bowel cancer screening program. In case of FIT positivity, subsequent colonoscopy was provided. 12 From the start of the program in 2014 until January 1 st 2021, 11,551,878 invitations were sent and 8,584,961 participants were adequately screened (74.3% uptake of primary screening test, all rounds together). Approximately 4.9% ( n  = 418,129) of the screened participants had an abnormal FIT result. There was an 82.5% uptake of endoscopy ( n  = 344,917). This resulted in 265,628 pathology reports (77% of all colonoscopies), with an average of 3.6 separate diagnoses per case (SD 2.4, median 3). All histopathological diagnoses from colorectal biopsies in the setting of the Dutch population screening program between 2014–2021 were retrieved from the Dutch nationwide pathology databank, PALGA, 13 registered under LZV‐2020‐116. This study was considered exempt from ethical approval due to anonymous data. We included all diagnoses that did not belong to the primary endpoints of the population screening program (i.e. colorectal cancer, conventional adenomas, hyperplastic polyps, sessile serrated lesions, and traditional serrated adenomas were all excluded). The flowchart is shown in Figure  1 . Exclusion criteria were lack of tissue, no diagnosis, no abnormalities, anal pathology, reactive changes, lymphoid hyperplasia, ischaemia, haematoma and haemangioma, pseudomelanosis, infections, amyloidosis, and endometriosis. Miscoded cases were those that were filed under the “other” category but belonged to the primary endpoints of the program (i.e. serrated polyps, adenomas, and adenocarcinomas). Flowchart of case selections. The results are described as frequency, linked to location and size, as well as to clinical parameters such as gender and age. When possible, incidence is given as number of cases per 10,000 colonoscopies. No additional statistics were performed. In the Results section, comparison with the existing literature is made for incidence and clinicopathological features.

From our original 63,263 included cases, we excluded 54,216 cases according to the exclusion criteria (Figure  1 ), leaving 9,047 cases for further analysis. This translates to a prevalence of 262 other polyps and malignancies per 10,000 colonoscopies. The most common specific findings are summarised in Table  1 . Clinical data per polyp diagnosis N : Number, NOS not otherwise specified. age range is 55–75 according to inclusion for population screening. Size in cm, age in years. The most common polyps were inflammatory polyps (Figure  2A ), over 65% of cases ( n  = 5950, 173 per 10,000 colonoscopies). These were not directly associated with the presence of IBD in other biopsies from the same patients. However, no clear definition of this type of polyp is available, resulting in a wide variety of histological morphologies. Distribution of polyp diagnoses. A : Relative frequency of the different diagnosis on a per‐polyp basis. B: Distribution according to location in the bowel. Inflammatory fibroid polyps are a distinct entity, with a well‐described morphology and accompanying molecular aberrations, and should be classified as a mesenchymal polyp. These most frequently present in the stomach or small bowel. 14 Only one case was diagnosed in our cohort (Table  2 ). No PDGFRA mutation was present, but the diagnosis was confirmed in the regional soft tissue expert panel. Summary of very rare diagnoses on a per‐patient basis Prolapse‐related polyps were present in 12 per 10,000 colonoscopies ( n  = 400). These were located mainly in the sigmoid (44%) and rectum (46%) (Figures  2B and 3I ), the former suggesting a relation with diverticular disease and the latter likely due to a multiplicity of causes, including mucosal prolapse and local trauma (obstipation, straining). A: Leiomyoma, B: lipoma, C: Perineurioma, D: ganglioneuroma, E: neurofibroma, F: Schwann‐cell hamartoma, G: juvenile polyp with dysplasia, H: Peutz–Jeghers polyp, I: prolapse‐related polyp, J: xanthoma, K: mantle cell lymphoma, L: MALT lymphoma, M: neuroendocrine tumour, N: neuroendocrine carcinoma, O: Metastasis of cutaneous melanoma, P: Metastasis of gastric cancer. Twenty‐seven mainly small leiomyomas were detected per 10,000 colonoscopies, most frequently in the sigmoid colon (Table  1 , Figure  2B , and Figure  3A ). In 12 patients, two leiomyomas were present, and in 147 cases this was the only diagnosis in the patient. They are benign proliferations derived from the muscularis mucosae, and are negative for gastrointestinal stromal tumour‐specific immunohistochemistry. 15 In line with the literature, 15 our cases were most common in the rectosigmoid and rectum. Sixteen lipomas were detected per 10,000 colonoscopies, most frequently in the caecum and ascending colon, in line with the literature 16 (Table  1 , Figure  2B , and Figure  3B ). This might be an underestimation, since these can be recognised by their endoscopic features and will not be sent for diagnosis. One case was diagnosed as an angiolipoma and another case as an intramucosal lipoma. Larger lipomas (over 4 cm in size) were present in the sigmoid ( n  = 4) and transverse colon ( n  = 3). In 73 patients, lipoma was the only diagnosis; the majority of those were small. A specific subgroup in the mesenchymal polyp category is formed by the neural polyps. The most common neural polyp is perineurioma (Figure  3C ), with 3 per 10,000 colonoscopies. The coincidence of perineuriomas with sessile serrated lesions 17 might be responsible for inflated numbers, as in particular the smaller perineuriomas often present with serrated crypts. The original title, benign fibroblastic polyp, was used 29 times. Ganglioneuromas were detected in 2 per 10,000 colonoscopies, most frequently in the sigmoid colon (Figure  2B , Table  1 , and Figure  3D ). Seven patients presented with multiple ganglioneuromas, ranging from 2 to 11. In all seven patients a genetic predisposition was suggested in the conclusion of the report. Neurofibroma (1 per 10,000 colonoscopies, Figure  3E ), Schwann‐cell hamartoma (0.6 per 10,000 colonoscopies, Figure  3F ), and granular cell tumour (0.4 per 10,000 colonoscopies) are very rare. Juvenile(‐like) polyps were the most frequently diagnosed hamartomas (12 per 10,000 colonoscopies, n  = 438). In 19, polyps dysplasia was present. Twelve patients presented with two juvenile polyps, one of them also had a dysplastic juvenile polyp (Figure  3G ). Juvenile polyps occurred most frequently in the sigmoid (Figure  2B ), just like the juvenile polyps with dysplasia (11/19). Peutz–Jeghers(‐like) polyps were diagnosed in 1 in 10,000 colonoscopies. On average, these were the largest polyps, with a mean size of 13 mm (Table  1 , Figure  3H ). There was also a rather large group of hamartomas that could not be further classified (1.7 per 10,000 colonoscopies); five patients with two hamartomas, one with three, and one with eight hamartomas. Small xanthomas (Figure  3J ) were mainly diagnosed in the rectosigmoid, with an incidence of 1 in 10,000 colonoscopies. In six cases this was the only diagnosis. Heterotopias were diagnosed very rarely (Table  2 ), and were located mainly in the rectum, in line with the literature. 18 The most common haematological malignancies (Table  3 ) were mantle cell lymphoma (Figure  3K ) and MALT lymphoma (Figure  3L ), with an incidence of 0.7 and 0.6 per 10,000 colonoscopies, respectively. In particular, mantle cell lymphoma presented as polyposis, in line with the literature. 19 Summary of malignant diagnoses per patient CLL‐SLL, chronic lymphocytic leukaemia/small lymphocytic lymphoma; EBV‐B‐LPD, Epstein–Barr virus‐associated lymphoproliferative diseases, B cell type; GIST, gastrointestinal stromal tumour; MALT, mucosa‐associated lymphoid tissue; NEC, neuroendocrine carcinoma; NET, neuroendocrine tumour. Distinction between direct invasion and metastasis was not clear on biopsy. Neuroendocrine tumours (NET, also including neuroendocrine carcinomas in this paragraph, Table  3 , Figure  3M,N ) presented in 6 per 10,000 colonoscopies, which is similar to the incidence in the English population screening program. 20 Most of these were grade 1 NET (82%), and most were located in the rectum (58%). In 33 patients metastatic disease (Table  3 , Figure  3O,P ) was present in the colon (1 per 10,000 colonoscopies). However, in a number of cases it was not entirely clear during colonoscopy whether direct ingrowth in the bowel wall was present. In at least five patients the primary cancer was already known. In line with the literature, gynaecological metastases were the most common. 21 In four patients, the origin of the metastasis was not clear based on histology alone. The other malignancies were three gastrointestinal stromal cell tumours, all low risk. One case with an atypical mesenchymal proliferation in the caecum was, after external consultation, booked out as a low‐grade leiomyosarcoma in a 70‐year‐old female and one Kaposi sarcoma was diagnosed in a 55‐year‐old male. This patient also presented with a high‐grade diffuse large B‐cell lymphoma. We do not have information on his HIV status.

This study provides an up‐to‐date overview of the incidence of relatively a rare colorectal diagnoses in a homogeneous group of asymptomatic patients within the setting of FIT‐based population screening. For most of these entities, these kinds of data are not or rarely available, or derived from historical large autopsy series. With the increased technical possibilities in modern endoscopy, more and smaller lesions are detected, necessitating information derived from a recent large patient series. Analysis of population screening cohorts, using structured reports from our national database, provides reliable data in large, relatively homogeneous cohorts. In the Results section, we already provided some background references linked to the specific diagnoses. In general, by the detection of lesions in asymptomatic patients, one could wonder about the clinical impact of our diagnoses. First of all, it is important to note that most of these lesions in general do not warrant follow up. Second, the detection of malignancies other than CRC has a direct impact for patients and their treatment. While the majority of NETs are adequately treated by polypectomy, and chronic lymphocytic leukaemia might not need treatment in all cases, other patients might have received their treatment earlier due to detection in the population screening program. Finally, there is a small group of patients in whom their hamartomatous polyps might point towards the presence of a hereditary cancer syndrome. For the diagnosis of juvenile polyposis, more than three juvenile polyps are necessary. 10 In 12 patients, two juvenile polyps were diagnosed. So, based on the histological diagnosis alone, these patients were not yet diagnosed with juvenile polyposis. The presence of dysplasia in a juvenile polyp is not part of the definition, but dysplasia is not described in sporadic juvenile polyps. 22 Additional staining for SMAD4 expression might also be helpful in the determination of a syndromic origin of juvenile polyps. 23 Patients with Peutz–Jeghers polyps, as well as patients with hamartomas that cannot be further classified are also candidates for a more thorough workup for a potential hereditary polyposis syndrome. For the former, the characteristic mucocutaneous pigmentation might be the key to the diagnosis of Peutz–Jeghers syndrome. Nonclassifiable hamartomas can be part of several syndromes, but are particularly often present in Cowden syndrome. 24 In the context of this syndrome, a wide variety of lesions can be observed, including lipomas and lymphoid polyps. The presence of multiple ganglioneuromas, as observed in seven of our patients, is considered a major criterion for this diagnosis. 25 This was well considered by the reporting pathologists, as evident by additional remarks in the conclusion of their reports, suggesting genetic counselling. However, the distinction between postinflammatory polyps and hamartomatous polyps might be impossible. One could argue that, based on the age of the screening population, the chances of discovering a hereditary cancer syndrome (i.e. juvenile polyposis or Peutz–Jeghers syndrome) are low. Indeed, the majority of patients with Peutz–Jeghers syndrome are diagnosed before their thirties. 26 Careful review can distinguish between Peutz–Jeghers polyps and mimics, 27 but true sporadic Peutz–Jeghers polyps are extremely rare. Additional clinical information is essential for the final diagnosis. The overlap between juvenile polyps and inflammatory polyps is evident, and some pathologists do not diagnose juvenile polyps in older patients, but refer to “inflammatory polyp, juvenile type” instead. 28 For juvenile polyposis, the mean age of diagnosis is 25 years, with a wide age range (0–86 years). 29 However, juvenile polyps can also occur as part of Cowden syndrome, which is more difficult to diagnose. 30 In order to distinguish simple postinflammatory polyps from these specific hamartomas, careful histological review in combination with clinical examination and (family) history is necessary. It might well be that most cases would be better diagnosed as postinflammatory polyps. Another potential pitfall is the recognition of prolapse‐related polyps. In particular in the rectum, these might be misdiagnosed as serrated polyps. 31 , 32 , 33 While prolapse‐related polyps fall into the spectrum of solitary ulcer syndrome, these are characterised by elongated and distorted crypts, with hyperplastic features, surrounded by proliferation of smooth muscle fibres from the muscularis mucosa. The presence of these stromal changes should prevent the diagnosis of a sessile serrated lesion, 31 which can be supported by the lack of BRAF mutations. 32 It is well known that colorectal cancer and precursors can be detected by (microscopic) blood in the stool. For other malignancies and the larger polyps, blood loss can also be expected. In addition, most patients with other polyps presented with colorectal cancer or precursors as well. However, small numbers of patients present with only minute lipomas or xanthomas (no other synchronous lesions present) that are most probably not the cause of a positive FIT. Potential causes include the coexistence of haemorrhoids, medications, 34 as well as, rarely, upper gastrointestinal cancers. 35 However, the incidence of the latter is too low to warrant standard workup of these patients. In this study, we present a timely overview of relatively rare diagnosis on colorectal biopsies and polypectomies in the setting of a national FIT‐based CRC screening program to provide incidence data and speculate about clinical consequences.

The worldwide implementation of bowel cancer screening programs has led to a steep increase in the number of colorectal biopsies and polypectomies. Not only has there been an initial increase in the number of colorectal cancer (CRC) cases, also numerous precursor lesions have been diagnosed. The increased frequency of these types of diagnoses in combination with several quality improvement strategies 1 , 2 , 3 and increased awareness of pathologists has led to an overall improved quality of diagnosis as well as an increased interest in gastrointestinal pathology. Datasets for standardised reporting have been developed and can be used to monitor trends in diagnosis. 4 Apart from CRC and its precursors (conventional adenomas and serrated polyps), within the population screening other diagnoses are frequently encountered. Inflammation is present in 4–5% of endoscopies within population screening programs. 5 Detection of asymptomatic inflammatory bowel disease (IBD) varied between 0.4 and 50 patients per 10,000 screened persons. 6 , 7 , 8 This development led to intense discussions about clinical management of these asymptomatic patients. However, several other diagnoses within the screening programs might have clinical consequences as well. Detection of malignancies other than CRC, but also hamartomatous polyps, might require action, since these may be indicators of hereditary cancer syndromes. 9 , 10 , 11 The incidence of these additional findings is not yet known. In the current study, we aimed to gather information on infrequent diagnoses in endoscopic specimens. We selected all histopathological diagnoses within the Dutch bowel cancer screening program and describe the relative incidence in asymptomatic patients and add characteristics of these other diagnoses.

No conflict of interest declared