This is not Lynch syndrome: lessons from misattributed diagnoses in constitutional mismatch repair deficiency

Abstract

Background

Underdiagnosis of constitutional mismatch repair deficiency (CMMRD) syndrome leads to suboptimal cancer surveillance and management of CMMRD patients. Assessing pitfalls that led to the misdiagnosis of CMMRD is important to improve care trajectories, and to highlight the importance of accurate molecular and pathology-based assessment of patients presenting with CMMRD-associated features.

Materials and methods

A retrospective chart review of two patients with molecularly confirmed CMMRD ascertained through the Medical Genetics service of the McGill University Health Centre (MUHC) was conducted to study the pathway and pitfalls to diagnosis. Records were reviewed and summarized as timelines to depict important events relating to diagnosis and management of CMMRD patients.

Results

Unfamiliarity with CMMRD contributed to a diagnosis delay and initiation of CMMRD-specific surveillance. Pitfalls along the diagnostic pathway included inaccurate clinical information relayed to pathologists, unfamiliarity with CMMRD-defining features on immunohistochemistry (IHC) analyses, IHC variability and unreliability, and lack of awareness of the pivotal role for medical genetics in the diagnosis of CMMRD.

Conclusions

Improved awareness of CMMRD in patients presenting with CMMRD-associated features can help guide IHC analysis and expedite referral to medical genetics for accurate molecular diagnosis. Consequently, timely CMMRD diagnosis improves surveillance and patient management and allows for appropriate genetic counseling for family members.

Highlights

• •CMMRD misdiagnoses are attributed to unfamiliarity with CMMRD and the genetic difference between CMMRD and Lynch syndrome.
• •Subjective interpretation of immunohistochemical staining alone cannot distinguish CMMRD and Lynch syndromes.
• •Misdiagnosis of CMMRD delays appropriate CMMRD surveillance and treatment.

Introduction

The term ‘Lynch syndrome’ (LS) was adopted to describe individuals with heterozygous germline pathogenic variants (GPVs) in one of the mismatch repair (MMR) genes: MLH1, MSH2, MSH6, and PMS2.¹ GPVs in the MMR genes have been shown to predispose patients to two hereditary cancer syndromes with some overlapping features. The underlying cause of LS is inheritance of a monoallelic GPV in an MMR gene.² When a second, somatic (tumor) ‘hit’ to the wild-type allele of the MMR gene occurs, DNA fidelity during replication is impaired, leading to persistence of naturally occurring DNA mismatches. This leads to the development of characteristic adult-onset tumors including colorectal, endometrial, and ovarian carcinomas.² In contrast, inheritance of biallelic GPVs in an MMR gene results in an earlier-onset cancer predisposition syndrome known as constitutional mismatch repair deficiency (CMMRD).³ The clinical presentation of CMMRD is variable with typical onset in childhood or young adulthood. The most common associated malignancies include lymphoma, early-onset gastrointestinal tumors, and brain tumors. A small subset of individuals may present with distinct clinical stigmata suggestive of neurofibromatosis type 1 with no history of malignancy.⁴ While LS is inherited as an autosomal dominant trait, CMMRD is autosomal recessive in its inheritance pattern. As both parents of an index case with CMMRD have co-occurring LS, CMMRD diagnosis has far-reaching implications for at-risk family members.

Variable MMR gene penetrance and patient phenotype⁵ can complicate the recognition and timely diagnosis of LS, necessitating universal screening of all colorectal and endometrial cancer tumors by immunohistochemistry (IHC) and microsatellite instability (MSI).⁶ However, reliance on IHC findings alone is not sufficient to distinguish MMR gene-related conditions in the absence of germline testing.^7,8 Lack of referral to genetic counseling has been associated with underdiagnosis of patients with LS (without mention of CMMRD).⁹

Efforts to improve timely diagnosis of patients with CMMRD have included the creation of a scoring system by the European Consortium Care for CMMRD (C4CMMRD) in 2014 to identify individuals who should undergo germline testing.³ More recently, the two largest CMMRD consortia established a robust pathway for the genetic diagnosis of CMMRD in relation to molecular diagnosis, ancillary testing, and clinical manifestations.¹⁰ In this study, we discuss the diagnostic pathways of two patients and highlight learning points to improve timely diagnosis of CMMRD.

Materials and methods

Both patient medical records were reviewed through the McGill University Health Centre (MUHC) electronic medical record. Timelines were made using Illustrator 2021 (Adobe Inc, San Jose, CA). After reviewing the content of this study, the two patients in this study provided their consent.

Results

Patient 1

A female patient developed T-cell lymphoblastic lymphoma in the first 5 years of life (Figure 1A) and was then treated with chemotherapy. Following disease relapse 5 years after initial diagnosis, the patient underwent reinduction chemotherapy followed by allogeneic stem cell transplant (cord blood, conditioning regimen included total body irradiation) at 11 years of age. The patient then had a resection of a perinasal keratoacanthoma, on which no IHC analyses were carried out at the time.

Figure 1.

Patient timelines. (A) Patient 1 [PMS2 Trans c.1021delA/delExon10, p. (Arg341Glyfs)] and (B) patient 2 [PMS2 c.2117delA, p. (Lys7072Serfs∗19)] timelines. The red and green timeline shades indicate time before and after constitutional mismatch repair deficiency syndrome (CMMRD),³ respectively. The dashed line does not follow time scale and indicates either no interaction with patient or no medical records available. The red font color indicates developments relating to CMMRD diagnosis. The blue color indicates surgical procedures. The timeline scale is indicated in orange color. CMMRD, constitutional mismatch repair deficiency; CRC, colorectal cancer; MMR, mismatch repair; MSI, microsatellite instability.

Nine years later, the patient presented with fever and generalized weakness and cross-sectional imaging demonstrated uptake in the rectosigmoid and transverse colon and metastasis to the liver. Colonoscopy confirmed the presence of a tumor in the sigmoid colon and six synchronous polyps. Pathology revealed moderately differentiated invasive adenocarcinoma with loss of PMS2 nuclear staining by in tumor cells, with intact PMS2 staining in the nuclei of nearby normal colonic mucosa. Ampliseq focus next-generation sequencing (NGS) panel testing of tumor DNA did not reveal any clinically relevant variants. The patient was started on pembrolizumab immune checkpoint inhibitor therapy.

The patient was offered multi-gene panel germline sequencing of tumor DNA which revealed compound heterozygosity for two pathogenic variants (PVs) in the PMS2 gene, c.1021delA p. (Arg341Glyfs) and an exon 10 deletion, respectively. Parental testing confirmed the variants to be inherited in trans. Review of family history for patient 1 showed that the parents are non-consanguineous. Further, the paternal side is non-contributory, except for the paternal grandmother who had breast cancer in her 70s. On the maternal side, an uncle had an unspecified cancer at age 39 years. No additional family history of LS-associated cancer was reported up to third-degree relatives.

Re-examination of the previously resected keratoacanthoma sample showed loss of PMS2 staining in epithelial and endothelial cells. In parallel, a re-review of IHC testing on the biopsy tissue was requested, which confirmed that the previously reported intact staining of PMS2 protein in normal mucosa was due to preserved IHC nuclear staining of PMS2 within circulating lymphocytes from the patient’s stem cell transplant donor. One month later, the patient underwent a total colectomy with ileorectal anastomosis and a partial left hepatectomy (segments 4b, 2, 3, and partial 5). Three invasive adenocarcinomas were noted along with 10 synchronous tubular adenomas.

Pembrolizumab treatment was continued for 2 years with a complete radiologic response. The patient’s cancer surveillance was also changed as per the International Replication Repair Deficiency Consortium (IRRDC) recommendations to include a brain magnetic resonance imaging (MRI) every 6 months, an annual whole-body MRI, abdominal ultrasounds, and a complete blood count every 6 months.¹¹ Despite the unequivocal documented evidence in support of a CMMRD diagnosis, review of the patient’s medical and surgical notes reveals misattributed diagnoses of ‘LS-like condition’ and ‘suspected LS’ in addition to CMMRD.

Patient 2

A male patient currently in his mid-30s was diagnosed as a teenager with colorectal cancer (CRC) in 2001 (Figure 1B) and underwent total colectomy with ileorectal anastomosis. Pathological analysis revealed a poorly differentiated medullary carcinoma. IHC testing revealed intact nuclear staining of MLH1 MSH2 and MSH6 proteins. PMS2 staining was not done as PMS2 involvement in MMR-related conditions was not known at the time. Despite no molecular diagnosis, the patient was attributed a working diagnosis of ‘suspected LS’. He received adjuvant 5-fluorouracil with methotrexate and was maintained on FOLFOX chemotherapy for 6 months. He continued to undergo cancer surveillance with sigmoidoscopy and/or endoscopy every 6 months for 4 years.

Six years later, the patient presented for an elective sigmoidoscopy as part of routine surveillance with resection of a rectal tubulovillous adenoma. A series of polypectomies ensued over the next several years resulting in clinical suspicion of an undefined ‘hereditary polyposis syndrome’ (Figure 1B). The patient was then advised to undergo laparoscopic (completion) proctectomy via a trans-anal total mesorectal excision with a J-pouch reconstruction, and a diverting loop ileostomy.

IHC analysis of samples from the final polypectomy before surgery demonstrated a change in IHC results with MLH1 and MSH2 intact staining, but loss of MSH6 nuclear staining and with intact positive internal control, raising the possibility of LS. In the same analysis, PMS2 staining was deemed ‘inconclusive’. While still under the banner of LS, analysis on samples collected intraoperatively revealed MSI and another set of varying IHC results. In this analysis, the report concluded on a weak nuclear staining of MLH1, normal staining of MSH2 and MSH6, and loss of PMS2 nuclear staining. This report made no reference to ‘internal controls’ or comments about staining in normal tissue.

The patient was then referred to medical genetics at the treating hospital. Family history examination showed no parental consanguinity. There were also no reports of paternal family history of malignancy and no maternal family history of MMR-associated cancers in first-degree relatives. However, a maternal grandmother had bladder cancer in her 70s. Multi-gene MiSeq panel (Illumina, San Diego, CA) testing of germline DNA revealed homozygous PVs in PMS2 c.2117delA (p. Lys706Serfs∗19), which has previously been reported in a child with early-onset CRC and French-Canadian ancestry.¹² The patient’s diagnosis was changed from LS to CMMRD, and the patient was initiated on cancer surveillance tailored to CMMRD.¹¹

Discussion

Unfamiliarity with CMMRD and the importance of genetic testing

Unfamiliarity with CMMRD as an MMR-related condition can result in misinterpreted IHC results without necessary clinical context. This is particularly important given that in most populations, CMMRD is an extremely rare diagnosis whereas LS is relatively common. In addition, variable IHC reliability may further complicate the ability of the pathologist to interpret the results. The Evaluation of Genomic Application in Practice and Prevention¹³ working group recommends testing all CRC cases for LS with IHC and MSI before germline sequencing.¹⁴ However, as the patient timelines (Figure 1A and B) suggest, IHC testing was not promptly followed by germline genetic testing even despite clinical suspicion of LS or another hereditary cancer syndrome. Therefore, a robust diagnostic pathway that includes genetic testing informed by a patient’s unique clinical presentation is necessary to allow for timely diagnosis of CMMRD. Misattributed diagnoses can lead to delays in appropriate cancer treatment and management, implementation of targeted cancer surveillance into adulthood, as well as genetic counseling and cascade testing of at-risk family members.

IHC findings are an essential component to the CMMRD diagnostic pathway. Loss of one or more of the MMR proteins in the nuclei of both normal and tumor tissues should raise the question of biallelic MMR deficiency. Through our review, loss of PMS2 staining in both normal and tumor tissues was usually regarded as a ‘technical failure’ or an ‘inconclusive result’. Also, in the case of patient 1, lack of clinical information provided to the pathologist regarding the patient’s prior stem cell transplant led to misattributed intact staining of PMS2 in background donor lymphocytes.¹⁵ While technical challenges can occur and a re-analysis is sometimes warranted, CMMRD must be considered as the most likely diagnosis after revealing the loss of MMR nuclear protein staining in all cells. In patient 2, multiple rounds of IHC analyses revealed different results, which alludes to the subjective interpretation of IHC results and the inherent limitations of IHC analysis. Therefore, familiarity of the clinical and pathological manifestations of CMMRD is needed for treating physicians and pathologists alike to understand the limitations of IHC results in the context of MMR deficiency-related conditions and complement IHC findings with genetic sequencing necessary for CMMRD diagnosis confirmation.

‘Syndrome’ diagnoses require genetic testing

While there are clinical manifestations associated with MMR deficiency syndromes, the terms ‘CMMRD syndrome’ and ‘Lynch syndrome’ are not clinical diagnoses. For example, a pediatric patient with CRC and who has a score ≥3 on the C4CMMRD scoring system³ does not have CMMRD until genetic testing confirms the diagnosis. However, even genetic testing may not be conclusive in CMMRD diagnosis. For example, not all biallelic combinations of variants can inform clinical decision making in CMMRD patients.¹⁶ Only pathogenic and likely pathogenic variants are considered actionable variants. The clinical value of variants of uncertain significance (VUS), which comprise 49% of the variants in the PMS2 gene,¹⁷ is yet to be determined. Interpreting genetic variants in the precise context of patient phenotypes is extremely important, especially when it comes to genes associated with both pediatric and adult-onset disorders, and with high frequencies of VUSs.

The PMS2 challenge and limitations to genetic testing

The two patients in this case series carry PMS2 gene PVs. Fifty percent of CMMRD cases confirmed by molecular analysis have been associated with biallelic mutations in the PMS2 gene.¹⁵ Identification of PMS2 PVs is particularly challenging. Firstly, PMS2 gene testing was not introduced until 2009, 9 years after MLH1 and MSH2 testing was available,¹⁸ contributing to patient 2’s CMMRD diagnosis delay (Figure 1B). Furthermore, analysis of the PMS2 gene is complicated by the presence of 14 pseudogenes that overlap with exons 1-5 of the gene,¹⁹ and an additional 15th pseudogene at the 3′ end (PMS2CL),²⁰ necessitating multiple tactics to restrict sequencing of these regions during PMS2 analysis.²¹

The identification of PMS2 gene PVs may remain a challenge even when using targeted NGS.²² Paralogous sequence variants (PSVs), which can be exploited to differentiate between the PMS2 gene and PMS2CL, can be exchanged between these two loci,^20,23 complicating the interpretation of NGS data. Nonetheless, bioinformatic workflows that aligned gene and pseudogene variants against the PMS2 gene reference sequence identified invariant PSVs that are useful in tracing the origin of variants observed in NGS reads.^20,24 Further, leveraging a streamlined NGS capture-based method followed by long-range PCR allowed for accurate screening of copy number variants in the 3′ PMS2 portion.²⁵ Notably in this method, copy ratios with exon deviations more specific to either PMS2 or PMS2CL were analyzed by pathologists blinded to the mutational status in algorithms with tested generalizability.²⁵ Together, these targeted clinically applicable and high-throughput genomics approaches improve PMS2 variant detection and harness the diagnosis of MMR-related syndromes.

Conclusion

Challenges to the timely diagnosis of CMMRD are multifactorial. These challenges include limited awareness of CMMRD as a distinct syndrome with clinical overlap with LS, difficulty in interpreting IHC results by pathologists in the context of incomplete clinical information and/or technical challenges, misattributed diagnoses in patients historically labelled as having LS, as well as delayed involvement of medical genetics in the diagnostic work-up of patients with CMMRD. The frequency of PMS2 PVs is markedly higher in CMMRD when compared with LS²⁶ and their presence in a young patient with cancer can independently raise the suspicion for CMMRD. Cancer surveillance offered to CMMRD patients necessitates the distinction from the more broadly ascertained LS patient population. Close collaboration can provide a more robust testing pathway that overcomes some of the technical challenges to accurate CMMRD diagnosis. Finally, early diagnosis of CMMRD can avoid long-term misattributed diagnosis of LS or other hereditary cancer syndrome, enabling timely genetic counseling and testing of at-risk family members and initiation of appropriate CMMRD management.