Misclassification of a frequent loss of function variant from PMS2CL pseudogene as a PMS2 variant in Brazilian patients

Introduction

The use of Next-Generation Sequencing has significantly improved the accessibility and effectiveness of genetic diagnosis for patients with hereditary cancer predisposition syndromes. However, challenges arise in complex variants, such as errors found in classifying variants within genes that have homologous genomic regions or pseudogenes with sequences very similar to the original gene. These findings highlight the ongoing need for refinement in variant identification and classification methodologies.

PMS2 is a gene involved in DNA mismatch repair. Deficiencies in this gene are associated with Lynch Syndrome (LS), a condition related to increased risks for developing colorectal, endometrial, ovarian and other cancers. Genetic testing for this gene is challenging due to the existence of multiple pseudogenes [1, 2]. Fourteen pseudogenes have been identified and described as ψ1 to ψ14, overlapping with some or all of PMS2 exons 1 to 5 and varying in length. Additionally, there is the PMS2CL pseudogene (formerly known as ψ0) with high homology to the 3’ end of PMS2 in exons 9 and 11 to 15 [3, 4].

Prior research has highlighted the misclassification of the loss-of-function (LoF) variant c.2182_2184delACTinsG within the PMS2CL pseudogene, mistakenly ascribed to exon 13 of the PMS2 gene [5]. In this study, we describe a series of cases harboring this frequent LoF variant in the PMS2CL that has been incorrectly identified as a pathogenic variant in PMS2, with varying nomenclatures throughout the years. We confirmed the variant location in PMS2CL using different molecular techniques in several patients. Additionally, we discuss the importance of performing alternative methods to circumvent the NGS limitations in distinguishing similar regions shared between PMS2 and its pseudogenes and the relevance of increasing sequencing efforts across diverse populations. Finally, we empathize the pivotal role these endeavors play in ensuring precision in genetic diagnoses and mitigating disparities in genetic testing.

Methods

Results

The allegedly PMS2 variant received two different nomenclatures on previous testing: V1: c.2186_2187delTC; p.(Leu729Glnfs) - rs587779335, and V2: c.2182_2184delinsG (p.Thr728Alafs) - rs1554294508. We observed that the nomenclature of the variant changed in 2020, after a software update. Upon visual inspection of the alignments from patients with these variants, we confirmed that both nomenclatures represent the same variant (Figure 1A). The difference between the variants is that for V1, a single nucleotide variant (SNV) PMS2:c.2182A>G; p.(Thr78Ala) is called as a separated variant, while in V2 this variant is considered part of the delins event, as recommended by HGVS nomenclature.

• Download figure
• Open in new tab

Figure 1.

Analysis of PMS2 V1/V2 variants and reclassification as a PSM2CL variant. A. NGS reads alignment of multigene DNA panel sequencing, showing the PMS2 exon 13 region in IGV. The image depicts the bases involved in V1 and V2 nomenclatures. B. RNA sequencing for the PMS2 transcript and analysis by CLC Genomics software, showing absence of V1/V2. C. Depiction of gene-specific PCR (GSP) performed for PMS2 exon 13 and PMS2CL exon 4. Genomic coordinates (Hg19) correspond to amplicons location. Left alignment shows mapping of PMS2 GSP, demonstrating the absence of V1/V2, and right alignment shows mapping of PMS2CL GSP, demonstrating the presence of PMS2CL: n.1122_1124delinsG.

Among the 647 individuals evaluated, 12 (1.8%) patients carried the V1/V2 variant, exhibiting variant allele frequencies ranging from 15% to 34% (Table 1). These patients presented distinct cancer types, such as breast (8), colorectal (2) and gastric (1). Four patients had germline pathogenic variants detected in other cancer predisposing genes. The racial distribution of these patients was diverse, with three patients self-identified as Black (one having molecularly confirmed African ancestry), one as Brown and three as White.

We performed gene-specific PCR of all 12 cases and determined that V1/V2 is not present in the PMS2 gene. Moreover, sequencing analysis of exon 4 of the PMS2CL pseudogene unequivocally demonstrated the presence of this variant (Figure 1B). The corresponding correct nomenclature for this variant in PMS2CL is n.1122_1124delinsG. To further confirm that V1/V2 is not present in the PMS2 gene, we conducted an analysis of PMS2 transcripts by nested PCR in two patients, and in both cases no read containing V1/V2 variant was detected (Figure 1C).

Discussion

Here, we present data on a frequent PMS2CL African variant found in the Brazilian population, which is incorrectly classified as a PMS2 variant. We detected this variant with two distinct nomenclatures (PMS2:c.2182_2184delACTinsG and c.2186_2187delTC) in 1.8% of genetic tests performed in our center. Through gene-specific PCR and PMS2 transcript analysis, we established that both variants represent the same PMS2CL variant (n.1122_1124delinsG). Moreover, in our center 4 patients have received a genetic test report from external laboratories containing this variant mistakenly described as a PMS2 LoF variant, 3 of them describing the variant as VUS and one as pathogenic, highlighting the relevance of our investigation.

Currently, for the PMS2 gene, V1: c.2186_2187delTC is classified as “Uncertain significance reviewed by expert panel” and V2 as “Conflicting” in ClinVar, with several laboratories classifying both as “pathogenic.” The conflicting data for classifying both variants is related to the need of confirming the variant as a PMS2 variant. The annotation by the International Society for Gastrointestinal Hereditary Tumours (InSiGHT) expert group in Clinvar regarding this variant is “This variant is likely to come from pseudogene”. When no confirmation is performed, most likely this variant should be assigned to the PMS2CL pseudogene, making it not disease-causing [7].

The V1:c.2186_2187delTC has been reported in the literature in more than 12 articles, published between 1995 and 2019. Few articles described the variant in compound heterozygosity with another PMS2 variant in patients affected with Turcot Syndrome [8] or Constitutional Mismatch Repair Deficiencies (CMMRD) [9], indicating a true occurrence of the variant in PMS2. However, V1 has also been reported in individuals with other cancer types, including colorectal [1, 10], breast [11, 12] and prostate [13], and pseudogene interference was not ruled out in most of these studies. A recent pediatric study reported the finding of V1 in 2 patients with pilocytic astrocytoma, but after applying long-range PCR, they determined that the variant belonged to the pseudogene [14].

The V2: c.2182_2184delinsG is much less cited in the literature, with only 2 articles referring to the variant [1, 5]. The more recent report of this variant most likely reflects updates in variant calling algorithms that incorporated the PMS2:c.2182A>G SNV as part of the indel variant, as recommended by HGVS rules. While Guindalini [1] did not describe any sequential technique confirming the variant to PMS2, Chong et al showed that indeed the variant was located at PMS2CL in all 5 tested patients. In Clinvar, the variant is described as “Conflicting”, with 4 clinical laboratories classifying the variant as Pathogenic/Likely pathogenic and one as VUS. Based on the evidence outlined above, we believe that both V1 and V2 have been incorrectly assigned to PMS2 and mistakenly classified as pathogenic in several articles and that these variants should be classified as pathogenic only when unequivocal confirmed to be within the PMS2 gene.

Global genomic databases (ExAC and gnomAD) report an incidence of these variants ranging from 2.5% to 5.3% in the African population, while in ABraOM (a Brazilian genomic database) [15] V1 appears at a frequency of 1.3%. Brazil has a significant percentage of Afro-descendant population with an average African genetic ancestry of 12.7% [5, 16]. In our study, 57% (4/7) of the patients tested with the presence of the PMS2CL variant declared themselves as Black or Brown, and one had a 66% African genetic ancestry molecularly confirmed.

Genomics disparities represent a current challenge in clinical genetics [17]. It is well-documented that minorities and underrepresented populations often have more VUS detected in clinical genetic testing. Our report emphasizes this issue by highlighting a common African variant frequently misclassified as a VUS or as pathogenic. Our results also underscore the importance of addressing these misclassification errors caused by NGS’s inability to differentiate short regions that are very similar between PMS2 and its pseudogenes. The misclassification of variants in the PMS2 gene can lead to significant consequences in clinical genetics, with clinicians misinterpreting the genetic risk and potentially providing inappropriate management or surveillance recommendations. Moreover, the racial disparity in genomic analysis further exacerbates the challenges faced by minority populations in receiving accurate and timely genetic testing results, stressing the need for increasing the available genomic data from diverse populations.