Increased KRAS G12C Prevalence, High Tumor Mutational Burden, and Specific Mutational Signatures Are Associated With MUTYH Mutations: A Pan-Cancer Analysis

Abstract

The aim of this study was to determine the pan-cancer landscape of MUTYH alterations and the relationship between MUTYH mutations and potentially actionable biomarkers such as specific genomic alterations, tumor mutational burden, and mutational signatures. We used a large pan-cancer comprehensive genomic dataset from patients profiled (tissue next generation sequencing) during routine clinical care. Overall, 2.8% of 229 120 solid tumors had MUTYH alterations, of which 55% were predicted germline. Thirty tumor types had a 2% or greater MUTYH mutation rate. MUTYH-altered versus -WT cancers had significantly higher tumor mutational burden and more frequent alterations in KRAS G12C, but not in KRAS in general; these observations were statistically significant, especially in colorectal cancers. Across cancers, PD-L1 expression levels (immunohistochemistry) were not associated with MUTYH alteration status. In silico computation demonstrated that MUTYH mutational signatures are associated with higher levels of hydrophobicity (which may reflect higher immunogenicity of neoantigens) relative to several other signature types such as microsatellite instability. Survival of patients with MUTYH-altered versus -WT tumors was similar. In conclusion, comprehensive genomic profiling suggests that several features of MUTYH-altered cancers may be pharmacologically targetable. Drugs such as sotorasib (targeting KRAS G12C) and immune checkpoint inhibitors, targeting the increased mutational load and higher neo-antigen hydrophobicity/immunogenicity merit investigation in MUTYH-mutated malignancies.

Introduction

Exposure to carcinogens, including oxidizing agents, alkylating agents, DNA cross-linking agents, and radiation is capable of damaging DNA.¹ The MUTYH gene is located on the short arm of chromosome 1 (1p34.1) and is responsible for coding the MYH glycosylase enzyme. This enzyme is part of the DNA base excision repair system, which repairs oxidized DNA damage. 8-Oxoguanine (8-oxoG), a major oxidized form of guanine generated by reactive oxygen species, is highly mutagenic due to its mispairing with adenine.² The function of MUTYH is in preventing oxidative damage-induced mutations, such as GC>TA transversions. Accumulation of 8-oxoG in DNA causes somatic mutations, eventually leading to cancer and neurodegenerative diseases.^3,4

MUTYH-associated polyposis (MAP) is an autosomal recessive disorder characterized by colorectal adenomatous polyps that confers a very high risk of colorectal cancer (CRC), predisposition to duodenal adenomas and a modest increase in risk for non-CRC malignancies.⁵ Colorectal cancer cells in patients with MAP fail to repair mismatches induced by 8-oxoG and consequently harbor an excess of KRAS c.34G>T transversions and structural abnormalities such as fusions, a widely recognized hallmark of oxidative stress.^6-9 The association between MUTYH mutation in patients with MAP and the increased frequency of KRAS G12C mutation (NM_004985.5(KRAS):c.34G>T (p.Gly12Cys)) has become especially relevant since KRAS G12C is now a druggable target.¹⁰ The base excision repair system defect, such as that induced by MUTYH alterations, may correlate with increased mutational burden and a mutational signature (COSMIC SBS18 and SBS36) that reflects typical oxidative damage.^11-15

We encountered a young (38-year-old) woman in clinic with metastatic CRC, a MUTYH mutation, and an elevated tumor mutational burden (TMB), who attained a complete response to a regimen that included immune checkpoint blockade. Because of the potential importance of MUTYH pathogenic alterations in cancer pathogenesis, and the paucity of published data available about them, we investigated a database of over 200 000 tumor samples to elucidate the MUTYH landscape, and its relationship with specific alterations and immune checkpoint blockade biomarkers.

Methods

Comprehensive Genomic Profiling Cohort

This study consisted of a pan-solid tumor cohort comprising 229 120 patients who underwent comprehensive genomic profiling (CGP), as part of routine clinical care. CGP on formalin-fixed, paraffin-embedded (FFPE) tissue biopsy sections was performed using hybrid capture on up to 395 cancer-related genes and select introns from up to 34 genes frequently rearranged in cancer (Foundation Medicine, Inc.). Approval for this study, including a waiver of informed consent and a HIPAA waiver of authorization was obtained from the Western Institutional Review Board (Protocol No. 20152817).

Identification of Genomic Alterations and Biomarker Patterns

Different classes of gene alterations, including short variants (base substitutions and indels), copy number alterations, and rearrangement events were identified, as described previously.¹⁶ Variants were classified as known/likely pathogenic or as variants of unknown significance (VUS). For short variants of interest, somatic/germline status for mutations was computationally predicted without matched normal tissue; in the validation testing of 480 tumor-only predictions against matched normal specimens, accuracy was 95% for somatic and 99% for germline predictions.¹⁷ Further examination of biallelic/monoallelic status of MUTYH alterations was also performed using the SGZ algorithm, where available.¹⁷ Specifically, cases with a single heterozygous MUTYH alteration were classified as monoallelic. Cases with 2 or more MUTYH alterations, or a single short variant in MUTYH predicted to be under loss of heterozygosity, or a homozygous copy number deletion of MUTYH were classified as biallelic. TMB (mutations/megabase [mut/Mb]) was determined on 0.8-1.1 Mb of sequenced DNA.¹⁸ TMB was assessed in 3 bins: <10 mut/Mb, 10-20 mut/Mb, and ≥20 mut/Mb. Microsatellite instability (MSI) status was by examining homopolymer repeat loci.¹⁹ PD-L1 expression was determined by immunohistochemistry (IHC) performed on FFPE tissue sections and scored using DAKO PD-L1 IHC 22C3 pharmDx’s tumor proportion scoring (TPS) method. Positive cases were evaluated at 2 thresholds: ≥1% and ≥50%.

Investigation of Mutational Signatures

COSMIC mutational signatures were identified as previously described.^19-22 Signatures were estimated in samples with at least 10 assessable mutations. Signatures were considered dominant if the maximum signature score met the threshold of 0.4. Given our interest in MUTYH-altered tumors, we performed a deeper investigation of SBS18 and SBS36, signatures implicated in defective base excision repair, including DNA damage due to reactive oxygen species.

Detecting rare signatures such as SBS18 and SBS36, which are predicted to result in 0.1-1.0 mutations/Mb when present, is challenging from a targeted panel test and may not contribute enough mutations per sample to be detectable. To overcome this, we used a pooled sample approach. Briefly, all variants that were both predicted non-germline and VUS were pooled for this analysis. Comparisons were performed between the mutant group, defined as patients with a germline MUTYH alteration (gMUTYH), and the wildtype (WT) group, defined as patients with no identified MUTYH alteration. Since mutagenic signatures are typically considered to be additive, the difference between the 2 groups was used to understand the additional effect of MUTYH mutations. The contributions of the known COSMIC v3.2 signatures were obtained using the method outlined previously.²⁰ The number and percentage of mutations attributed to each signature were calculated relative to the total number of pooled mutations.

The stability of this pooled approach was evaluated by resampling the cohorts to a jackknife sample size, defined as the mutations from half the number of samples available in the smaller of the 2 cohorts being compared (mutant and wild type). Both the cohorts were resampled without replacement to the jackknife sample size 1000 times, and signature attribution was assessed on each of the resampled pools of mutations. This provides both the median and 95% CI for the contribution of all known COSMIC v3.2 mutation signatures in each cohort.

In Silico Mutagenesis Probabilistic Simulations

We applied an in silico hydrophobicity analysis method to understand the immunogenicity of the increased mutation load as a result of MUTYH mutation. Every possible combination of 2 codons, comprising 6 nucleotides, was generated and the Single Base Substitution (SBS) mutational signatures defined by COSMIC v3.2²⁰ for MUTYH (SBS18 and 36), MSI (SBS6, 14, 15, 20, 21, and 26), and APOBEC (SBS2 and 13) were applied to each, separately. The 6-nucleotide length allowed us to account for point mutations, contextually defined by their flanking nucleotides in COSMIC, in all possible reading frames of a codon, which requires at least 5 nucleotides. In total, 4096 di-codons, before and after mutation, were transformed into their matching amino acid pairs whose hydrophobicity was then additively quantified using the Kyte-Doolittle Hydrophobicity Scale²³ and weighted by their frequency on the human exome, starting from the base frequencies in Kazusa’s codon usage database²⁴ and updated at every iteration, and the probability of signature-defined mutagenesis on each 6-nucleotide fragment.

To illustrate this algorithm more concretely, consider the 6-nucleotide sequence TCTGAT which encodes the peptide Ser-Asp that has a Kyte-Doolittle hydrophobicity index of (−0.8) + (−3.5) = (−4.3) AU. The first codon at the second position can be mutated with SBS36’s, a MUTYH signature,²¹ most common mutation TCT > TAT, which has a frequency of occurrence of 0.208 according to data available at²⁵ resulting in the sequence TATGAT, or the dipeptide Tyr-Asp, which has a hydrophobicity index of (−1.3) + (−3.5) = (−4.8) AU, a difference of (−0.5) AU over the original. This difference in hydrophobicity is then multiplied by the joint probability of the codons’ occurrence and mutation, calculated by multiplying the frequency of the original dicodon in the human coding genome (0.0152 × 0.0218 = 3.31 × 10^-4) and the frequency of TCT > TAT mutation under SBS 36 (0.208), resulting in a joint mutation-dicodon probability of 6.89 × 10⁻⁵. Multiplying the change in hydrophobicity by the joint mutation-dicodon frequency yields a relative hydrophobicity change of 6.89 × 10^-5 × (−0.5) = (−3.45 × 10^-5). The same calculation was performed for every potential nucleotide substitution in all positions on the 6-nucleotide stretch with 2 flanking nucleotides (2nd through 5th positions, inclusive), on all 4096 combinations of 6 nucleotides. At the end of each round of mutagenesis, the frequency of codons is updated to reflect the outcome of the just-completed round, and the succeeding cycle is started using these updated codon frequencies.

Every iteration of mutagenesis is equivalent to an arbitrary unit (AU) increase in exposure to the signature’s mutagen, with multiple iterations corresponding to more frequent exposures. To model the long-term effects of mutagen exposure on proteome hydrophobicity, we performed these simulations for 100 iterations per signature (SBS18, 36, 6,14,15, 20, 21, 26, 2, and 13).^26,27

Overall Survival Analysis From a TCGA Pan-Cancer Cohort-In-Silico Analysis

To analyze the influence of alterations within the MUTYH gene on overall survival (OS) in patients belonging to a pan-cancer cohort, we queried the gene using the query function of cBioPortal web tool. Thirty two studies incorporating solid tumors and hematological malignancies from TCGA Pan Cancer Atlas containing 10 967 samples from 10 953 patients were used for plotting the Kaplan-Meier curve. Criteria for OS analysis, by default settings, include mutations, structural variants/fusions, and copy number alterations. The plot captures data from up to 370 months of survival. The MUTYH-altered group includes 169 patients, while the unaltered group has 10 634 patients. The P-value is calculated using the log-rank test, and q-value using the Benjamini-Hochberg FDR correction procedure.

Statistical Analysis and Software

Enrichment of MUTYH alterations in specific patient subpopulations was evaluated using a Fisher’s Exact test with false discovery rate (FDR)-based correction for multiple testing. Computation and plotting were carried out using Python 2.7 (Python Software Foundation) and R 3.6.1 (R Foundation for Statistical Computing).

Results

MUTYH Alterations are Observed Across Multiple Solid Tumors

In a pan-solid tumor cohort of 229 120 patients who underwent genomic profiling during routine clinical care, MUTYH alterations were detected in 6297 patients (2.8%). These MUTYH alterations were frequently predicted to be pathogenic, with 66% of the MUTYH-altered samples having at least one pathogenic MUTYH alteration (Fig. 1A; Supplementary Table S1). Pathogenic MUTYH alterations were identified in over 2% of the samples in 30 tumor types, including germ cell (3.6%), adrenal gland (3.1%), colorectal tumors (2.7%), and small intestine tumors (2.6%) (Fig. 1A). Across tumor types, a majority of the MUTYH alterations were predicted to be monoallelic (Fig. 1A). Additionally, we used a computational approach to predict the subset with predicted germline MUTYH alterations (gMUTYH).¹⁷ Among the MUTYH-altered samples, 55% harbored a predicted germline MUTYH alteration (prevalence of 1.5% across all solid tumors). Several tumor types were frequently positive for gMUTYH: germ cell (2.3%), salivary gland (2.2%), cholangiocarcinoma (1.9%), adenoid cystic carcinoma (1.9%), small intestine (1.8%), and colorectal cancers (1.8%) (Fig. 1A).

MUTYH alterations frequently comprised of single-nucleotide variants and short indels, with G382D (n = 2277) and Y165C (n = 867) being the most commonly identified alterations (Fig. 1B). In addition, recurrent splice site variants, such as 892-2A > G (n = 278), 1434 + 2C > T (n = 67) and 1145-2A > G (n = 23) were also frequently observed (Fig. 1B; Supplementary Table S2). Most alterations were missense and short indel alterations; copy number alterations and rearrangements impacting MUTYH were relatively rare (Supplementary Table S2).

MUTYH-Altered Samples Exhibit High Tumor Mutational Burden (TMB), But Not PDL1 Overexpression

Further, we examined the association between the presence of MUTYH alterations and TMB. MUTYH alterations showed a strong association with TMB-High (≥10 mut/Mb) status, and especially with TMB ≥ 20 mut/Mb (Fig. 2; Supplementary Fig. S1; Supplementary Tables S3 and S5) in the pan-cancer cohort, especially in colorectal cancer and adrenal tumors. MUTYH-altered cases had a significantly elevated TMB compared to wild-type (WT) samples (Wilcoxon, P < 10⁻⁵), with increased median (5 vs. 3.8 mutations/megabase, mut/Mb) and mean (14.9 vs. 7.2 mut/Mb) TMB in MUTYH-altered cases compared to WT samples. In comparison to all MUTYH-altered cases, samples with germline MUTYH (gMUTYH) alterations had a slightly lower TMB (median = 3.8 mut/Mb; mean = 7.5 mut/Mb). This pattern was consistently observed across multiple tumor types (Fig. 2A, Supplementary Table S3). Of note, among 517 adrenal gland tumors, cases with a MUTYH alteration (n = 18; 10 gMUTYH) had a significantly higher TMB than those that were WT (median 9.2 vs. 2.5 mut/Mb, P < 10^-4).

High PD-L1 expression and tumor mutational burden (TMB-H, ≥10 mut/Mb) have been widely used as biomarkers to identify patient subpopulations who would benefit from immune checkpoint inhibitors across multiple tumor types.^28-30 Having observed an elevation of TMB in MUTYH-altered samples, we then sought to characterize the association between these immune checkpoint blockade biomarkers and MUTYH alterations. We did not observe a significant association with PD-L1 expression (Fig. 2B; Fig. 2C; Supplementary Fig. S1) in the limited subset of samples with available PD-L1 (DAKO 22C3) expression (Supplementary Table S4).

MUTYH Alterations Show Specific Genomic Associations in Colorectal Cancers

We examined the prevalence and genomic association of MUTYH alterations in 31 624 CRC samples. Overall, MUTYH alterations were observed in 2.7% of CRCs, with gMUTYH observed in 1.8% of the overall cohort and in 65% of all MUTYH-altered CRCs (Supplementary Table S1). Similar to the pan-solid tumor cohort, G382D (n = 382) and Y165C (n = 124) were the most frequently identified MUTYH alterations (Supplementary Table S2). An additional 32 recurrent short variant alterations were identified in the cohort, including A371fs*23, R231H, R189C, and splice site alterations such as 892-2A > G, 1434 + 2C > T, and 1145-2A > G (Supplementary Table S2).

MUTYH-altered CRC tumors were more frequently TMB-H (≥10 mut/Mb) compared to MUTYH-WT CRC (15.7% vs. 7.1% respectively, P < 10⁻⁵) (Fig. 3A). Of note, among the 135 TMB-H MUTYH-altered CRCs, 85 (63.0%) had a TMB ≥ 20 mut/Mb. Statistical analysis revealed significant association between MUTYH and high TMB, especially with TMB ≥ 20 mut/Mb (odds ratio = 3.3, P < 10⁻⁵; Fig. 3B; Supplementary Table S5). Similarly, gMUTYH and pathogenic MUTYH alterations were also associated with high TMB (Fig. 3A; Supplementary Fig. S1; Supplementary Table S5). As with the overall pan-solid cohort, we did not observe any association with PD-L1 status in CRC (Fig. 3B; Supplementary Fig. S1; Supplementary Table S4).

Further examination based on MSI status revealed a higher prevalence of MUTYH alterations in MSI-H CRC compared to MSS CRC (7.4%, n = 75/1020 vs. 2.6%, n = 756/29 607 respectively, P < 10⁻⁵; Fig. 3C; Supplementary Table S1). The differences were less pronounced for gMUTYH (2.4% vs. 1.8%, P = .18; Fig. 3C; Supplementary Table S1) and pathogenic MUTYH alterations (2.8% vs. 2.0%, P = .09; Supplementary Table S1).

MUTYH Alterations Co-Occur With KRAS G12C Mutations

We next examined any association between KRAS alterations with MUTYH across different tumor types (Supplementary Table S6). Notably, while KRAS alterations overall were not associated with MUTYH status, KRAS G12C frequency was significantly higher in MUTYH tumors (5% in MUTYH-altered vs. 3% MUTYH-WT samples, P < 10⁻⁴) (Supplementary Table S6). In addition to being increased in the pan-solid tumor cohort, KRAS G12C was particularly increased in specific cancers such as colorectal cancers (Supplementary Table S6).

Increased transcriptional base mismatches, which develop against oxidative stresses as a result of MUTYH deficiency, lead to specific GC > TA transversions. These specific base mismatches can cause increased mutation rates and other co-occurring alterations.^11,12,31-34 To discern if MUTYH mutagenic processes were more likely to be associated with particular KRAS alterations, we evaluated the spectrum of KRAS alterations based on MUTYH alteration status.^33,34 In the overall CRC cohort, KRAS alterations were observed in nearly 50% of samples, with the prevalence being similar in MUTYH-altered samples: 50.2% in MUTYH, 56% in gMUTYH-altered samples (Supplementary Fig. S2; Supplementary Table S6; Fig. 3B). While the prevalence of KRAS was similar across all mutation subgroups, a closer examination of specific KRAS alleles revealed strong associations with KRAS G12C (Fig. 3B; Supplementary Table S6). KRAS G12C alterations were significantly increased in MUTYH-altered samples compared to WT (12.5% vs. 3.3% respectively; Odds ratio = 4.5, P < 10⁻⁵; Fig. 3B; Supplementary Table S6). Of note, this association was stronger in gMUTYH-altered CRC (odds ratio = 5.8, P < 10−⁵; Fig. 3B; Supplementary Table S6). The association of KRAS G12C was also observed in MSS CRC (Fig. 3D; Supplementary Table S6), whereas we were underpowered to test these patterns in the MSI-H cohort with only 8 out of the 1020 cases exhibiting KRAS G12C alterations.

Beyond CRC, other gastrointestinal tumors (eg, esophagus, stomach, small intestine) also showed the elevated frequency of KRAS G12C among MUTYH-altered samples in comparison to WT samples (Fig. 3E). In a similar assessment of KRAS G13D, we did not observe any association with MUTYH status (Supplementary Fig. S1; Supplementary Fig. 2; Supplementary Table S6).

MUTYH Alterations Show Specific Mutational Signatures With Elevated Immunogenicity Potential

MUTYH mutational signatures SBS18 and SBS36 have been implicated in defective base excision repair, including DNA damage due to reactive oxygen species.^33,34 Therefore, we sought to further characterize these signatures in our cohort of MUTYH-altered CRCs. However, targeted panel testing often poses a challenge in detecting these rare signatures. Notwithstanding this limitation, SBS18 and SBS36 combined were more prevalent in MUTYH-altered CRCs (prevalence 5.7% MUTYH vs. 0.2% WT, P < 10⁻⁵; 9.9% gMUTYH vs. 0.2% WT, P < 10⁻⁵; Fig. 4A; Supplementary Table S7). This pattern of elevated SBS18 and SBS36 burden was also found in samples with predicted monoallelic (2.5%) and biallelic (22.8%) pathogenic MUTYH alterations but not in MUTYH wild-type samples (0.2%) (Supplementary Fig. S3). To better assess mutational signature processes based on the MUTYH status, we then applied a pooled approach to characterize the contribution of each signature to the observed spectrum of alterations for each cohort (see Methods). Relative to the MUTYH WT samples, MUTYH-altered samples had a significantly larger number of alterations predicted to result from SBS18 and SBS36 (P < 10⁻⁵, relative mean compared to MUTYH WT: 5.6x and 4.4x mutations, respectively; Fig. 4B). This was consistently observed in a subset of gMUTYH samples as well as predicted pathogenic MUTYH-altered samples (Fig. 4B; Supplementary Fig. S3).

Hydrophobicity may reflect immunogenicity.³⁶ Therefore we further evaluated the impact of repeated in silico MUTYH-related mutagenesis over the estimated overall hydrophobicity of the coding genome The SBS36 MUTYH signature remains among the most hydrophobic signatures; SBS18 MUTYH signature exhibited moderate to high hydrophobicity, although not to the extent of SBS36 (Fig. 4C).

Survival Analysis Did Not Reveal Differences Based on MUTYH Status

Survival analysis revealed no statistically significant difference in overall survival (OS) between patients that were MUTYH-altered (n = 169) and those WT (n = 10 634) for MUTYH in a pan-cancer cohort (P = .220) (Fig. 5). The OS in the MUTYH altered group is 65.49 months (95% CI, 50.10-107.14) and in the unaltered group is 79.59 months (95% CI, 74.73-84.66). In addition, neither disease-free survivals (P = .890) nor progression-free survival (P = .931) was found to be statistically significant between MUTYH-altered and WT groups (data not shown).

Case Presentation

In February 2020, a 38-year-old woman was admitted to the emergency department with acute intestinal obstruction. Primary diversion colostomy was performed on the patient who was suspected of having a mass obstructing the lumen in the sigmoid colon, with 3 cm metastasis in segment 7 of the liver, and suspected of peritoneal involvement. Endoscopic biopsy identified adenocarcinoma, and the patient was initiated on systemic chemotherapy (leucovorin calcium [folinic acid], fluorouracil, and oxaliplatin: FOLFOX-6). Biopsies were obtained from the patient’s metastatic sites, particularly the liver, for confirmation prior to initiating therapy. These biopsies provided evidence of tumor presence and confirmed the stage IV disease status. Upfront tumor testing identified a PD-L1 tumor proportion score (TPS) of 80% (Dako 22C3 pharmDxTM IHC) and TMB by F1CDx was 16 mut/Mb (Supplementary Table S8 for complete genomic profile). The patient also exhibited a CD274 gene rearrangement and a KRAS G12C mutation. Importantly, she had a MUTYH alteration and an SBS18 mutational signature (consistent with MUTYH alteration). Nivolumab 240 mg every 2 weeks schedule was added to the FOLFOX6. After 4 months of chemo-immunotherapy, consolidative radiotherapy was given to the rectum, concomitantly with nivolumab and continued infusion of 5-fluorouracil. Afterward, radiotherapy, curative rectal surgery, and resection of the metastatic focus in the liver were performed. A complete pathological response was detected in the resected surgical specimens. The patient, who received maintenance nivolumab continues to be followed up and treated as no evidence of disease at 24 months after diagnosis. She has a strong family history of CRC affecting 2 first-degree relatives (Informed written consent was obtained from the patient to share the data).

Discussion

To evaluate the prevalence of MUTYH mutations, we used a large pan-cancer comprehensive genomic dataset from patients profiled during routine clinical care. Overall, 6297(2.8% of 229 120) solid cancers had MUTYH alterations, of which 55% were predicted germline. While the details of MUTYH-associated polyposis syndrome are known, the landscape of MUTYH mutation in cancers and its association with co-occurring mutations and genomic signatures have not been previously evaluated in a large-scale fashion.

Prior literature identified approximately 1%-2% of the general population with a single (monoallelic) germline MUTYH pathogenic variant, with 2 (biallelic) pathogenic variants identified in<1% of individuals diagnosed with CRC.^27,37 We detected MUTYH-related pathogenic changes in 6297 (2.8%) patients in the pan-solid tumor cohort (n = 229 120) with 30 tumor types harboring a MUTYH frequency of 2% or higher, and the highest rate of mutation in germ cell tumors and adrenocortical cancers. Among our MUTYH-altered samples, 55% showed a germline MUTYH alteration (gMUTYH); therefore gMUTYH was seen in ~1.5% of all solid tumors. Barreio et al also detected a 1.4% incidence of gMUTYH in gnomAD, which is consistent with our findings.³⁷

MUTYH-altered cases had a significantly elevated TMB compared to wildtype samples, across multiple tumor types in our study. Consistent with our observation, MUTYH-altered CRCs are also characterized by increased mutation and neoantigen burden, similar to patients with mismatch repair (MMR) defects. MUTYH alterations being associated with elevated mutation rates have been reported most frequently in CRC, but also in gastric, pancreatic, ovarian, and adrenocortical cancers.^3,15,29-31

Along with the increased mutational burden, PD-L1 IHC expression was also evaluated as one of the markers predicting the immune response.²⁷ Although Permata et al identified a relation between defects in the base excision repair (BER) system and increased PD-L1 expression,³⁸ relevant data in the literature is limited. No associations were found between the presence of PD-L1 expression and MUTYH alterations in our analysis.

It is well known that increased oxidative stress in the presence of loss of function MUTYH mutations leads to increased mutation load and specific GC > TA transversion. Therefore, some specific mutations may co-occur in this scenario. Mutations such as KRAS G12C and those in APC, and SMAD are most common.^11,12,31-34 In our patient set, KRAS mutations are observed at similar rates in MUTYH-altered compared to wildtype tumors; however, the KRAS G12C mutation is found at significantly higher rates in MUTYH-altered across cancers. This was especially prominent in patients with CRC, with a more pronounced difference in germline MUTYH CRC cases. Other gastrointestinal tumors (eg, esophagus, stomach, and small intestine) also showed elevated levels of KRAS G12C among MUTYH-altered samples in comparison to WT samples. Consistent with our observations, Salem et al recently reported that KRAS G12C-mutant CRCs have a higher prevalence of MUTYH mutations than non-KRAS G12C-mutant CRCs (5.77% vs. 2.28%, OR = 2.62, P = .005).³¹ Other studies also support this observation.^9,11,31-34 The 4-fold increased prevalence of KRAS G12C in MUTYH-mutated microsatellite stable CRCs is an important finding since KRAS G12C is being recognized as a druggable alteration, with the FDA approval of sotorasib in non-small cell lung cancer.³⁹

Somatic mutations in cancer genomes are caused by multiple mutational processes, each of which generates a characteristic mutational signature.²¹ In our dataset, we examined the mutational signature in cases with MUTYH mutations and found COSMIC SBS18 and SBS36 to be elevated in these patients compared to MUTYH-WT patients at both a per-specimen level and using a pooled approach to compare cohorts. These 2 signatures reflect DNA damage as a result of defects in the BER system and an increase in oxidative stress.^33,34 Presence of dominant SBS18 and SBS36 mutation signatures in these patients also reveals the role of oxidative stress and MUTYH mutations in CRC oncogenesis.^2,15,33,34

The success of immunotherapy in our index patient suggests that the defect in the BER system may be the result of increased mutational burden and neoantigens, similar to that seen in cancer with mismatch repair gene changes. Furthermore, several authors have indicated that MUTYH-altered tumors can also be immune-hot tumors with increased intra-tumor lymphocyte infiltration similar to mismatch repair-defective cases.^11,40,41

The relationship between higher hydrophobicity and immunogenicity, as well as the association between hydrophobicity and improved outcomes after immunotherapy, has been well-documented in previous studies.^{22,26,35,36,42} We found that the MUTYH signature is amongst the most hydrophobic of the signatures. Therefore, this observation provides further evidence that cancers harboring the MUTYH-altered signature may benefit from immune checkpoint blockade, in the same way seen in our index patient. This concept merits evaluation in prospective clinical trials.

There are studies showing that MUTYH-altered CRCs bear molecular and immune infiltrate structures similar to defective mismatch repair CRC cancers.^9,40,41 For this reason, we know that the prognosis is better and survival is longer in patients with mismatch repair CRC cancer. In a study that looked at MUTYH-altered versus -sporadic CRC survival, it was reported that the survival of MUTYH-altered CRC cases was better.⁴³ In contrast, we examined the survival of MUTYH-altered and -WT patient pan-cancer cohorts in the TCGA database; no survival difference was found. Due to the small number of cases, the CRC cohort could not be examined.

This study has several limitations. First, clinical outcomes and family/pedigree information were not available for the patients in this study. Second, germline testing from the patient’s normal tissue was not performed. We therefore had to rely on a computational algorithm for MUTYH germline prediction which is detailed described in the method section. Third, it is important to note that tumors with a higher tumor mutational burden may harbor a wider range of mutations, including those affecting genes such as MUTYH. Consequently, it becomes imperative to further investigate the immunotherapeutic implications specifically for tumors with MUTYH alterations.

In conclusion, in a dataset of 229 120 solid cancers, we found that 2.8% had MUTYH alterations, of which 55% were germline. Cancers with MUTYH alterations had higher TMB, more frequent alterations in KRAS G12C (a potentially actionable KRAS mutation) (both observations were especially pronounced in CRCs), but no increase in KRAS alterations in general, and a mutational signature with higher levels of hydrophobicity (reflecting immunogenicity) than most microsatellite unstable signatures. PD-L1 expression levels were however not increased. Survival of patients with MUTYH-altered versus -WT tumors did not differ. The finding of higher TMB and higher hydrophobicity indicate that patients bearing MUTYH-altered cancers may benefit from immune checkpoint blockade (as observed in our index patient), although the lack of association with higher PD-L1 expression indicates that the situation in individual tumors may vary and that defining whether or not checkpoints other than PD-1/PD-L1 are important in MUTYH-altered cancers warrants future studies. The role of sotorasib⁴⁴ (a KRAS G12C-targeted molecule, FDA-approved for lung cancer) also merits further investigation in MUTYH-mutated malignancies.

Acknowledgments

We thank Gunes Acar, MD, of Gunes Acar Danismanlik, Istanbul, for providing medical writing assistance.

Funding

Foundation Medicine, Inc. has provided the data to the investigators solely for purposes of the research. Funded in part by NIH grants U01CA217885, U24CA220341, P30CA023100, and U24CA248457 (TP).

Conflict of Interest

Umut Disel: Foundation Medicine as consultant, speaker fees, and/or advisory board. Smruthy Sivakumar: Employee at Foundation Medicine, a wholly owned subsidiary of Roche and a shareholder in Roche. Zoe Fleischmann: Employee at Foundation Medicine, a wholly owned subsidiary of Roche and a shareholder in Roche. Ethan S. Sokol: Employee at Foundation Medicine, a wholly owned subsidiary of Roche and a shareholder in Roche. Razelle Kurzrock: has received research funding from Biological Dynamics, Boehringer Ingelheim, Debiopharm, Foundation Medicine, Genentech, Grifols, Guardant, Incyte, Konica Minolta, Medimmune, Merck Serono, Omniseq, Pfizer, Sequenom, Takeda, and TopAlliance; as well as consultant and/or speaker fees and/or advisory board for Actuate Therapeutics, AstraZeneca, Bicara Therapeutics, Biological Dynamics, Daiichi Sankyo, Inc., EISAI, EOM Pharmaceuticals, Iylon, Merck, NeoGenomics, Neomed, Pfizer, Prosperdtx, Roche, TD2/Volastra, Turning Point Therapeutics, X-Biotech; has an equity interest in CureMatch Inc., CureMetrix, and IDbyDNA; serves on the Board of CureMatch and CureMetrix, and is a co-founder of CureMatch. The other authors indicated no financial relationships.

Author Contributions

Conception/design: U.D., E.S. Acquisition of the data: U.D., S.S., Z.F., A.I., T.P. Data analysis and interpretation: All authors. Manuscript writing: U.S., S.S., T.P., A.I. Final approval of manuscript: All authors. All authors made a substantial, direct, and intellectual contribution to the work and approved it for publication.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References