If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Mutations in the leucine-rich repeat kinase 2 (LRRK2) gene are the most common cause of familial Parkinson disease. Although LRRK2-related Parkinson disease patients have a heightened risk of certain nonskin cancers, including breast cancer, it is unknown whether LRRK2 somatic mutations occur and are associated with breast cancer. The objective of this study was to evaluate the occurrence of LRRK2 somatic mutations in breast cancer and the clinicopathologic features associated with LRRK2-mutated tumors. Using The Cancer Genome Atlas Breast Cancer Project, somatic LRRK2 DNA sequence information was obtained for 93 cases, of which 17 cases (18%) with 18 mutations were identified. LRRK2-mutated mammary carcinomas are enriched with stop-gain, truncating mutations predicted to result in loss of function; missense mutations frequently targeted the GTPase and kinase domains. Tumors displayed predominantly high-grade morphology with abundant granular eosinophilic cytoplasm, resembling mitochondria-rich apocrine-like carcinomas. Exploration of the genomic landscape of LRRK2-mutated carcinomas yielded frequent TP53 deactivation and a remarkably high tumor mutation burden. More important, breast cancers with LRRK2 mutations are associated with reduced patient survival compared with The Cancer Genome Atlas Breast Cancer Project cohort. These findings, for the first time, show that somatic LRRK2 mutations occur frequently in breast cancer, and the high mutation burden seen in this subset of tumors suggests that LRRK2 mutations may herald benefit from immune checkpoint inhibition.
Cancer and Parkinson disease (PD), the most common movement neurodegenerative disorder, at first glance seem pathophysiologically dissimilar. Although cancer results from an uncontrolled regenerative process, PD is characterized by neuronal cell death. Although these disease processes appear to be mechanistically contradictory, an association between PD and cancer has been established, suggesting overlap in the underlying biochemical dysfunction.
Despite this observation, the link between LRRK2 mutations and breast cancer has been largely unexplored. We report, for the first time, somatic LRRK2 mutations in breast cancer, and that this subset of tumors displays high-risk features with a high mutation burden, making them promising candidates for immune checkpoint therapy.
Materials and Methods
Clinical information and simple somatic mutations, including single-nucleotide variants and small insertions/deletions, from whole exome sequencing and copy number information available through The Cancer Genome Atlas (TCGA) Breast Cancer Project were downloaded from the Genomic Data Commons Data Portal. For an extensive description of the tumor variant annotation workflow performed by the Genomic Data Commons DNA-sequencing analysis pipeline, please see National Cancer Institute Genomic Data Commons (https://docs.gdc.cancer.gov/Data/Bioinformatics_Pipelines/DNA_Seq_Variant_Calling_Pipeline/#tumor-only-variant-annotation-workflow, last accessed July 10, 2020). The simple somatic mutations, single-nucleotide variants, and small insertions/deletions were subjected to the current guidelines by the American College of Medical Genetics. Briefly, LRRK2 nonsense, frameshift, and canonical splice-site sequence variants were considered pathologically significant because all of these types of mutations lead to either a stop codon or truncated, incomplete, or typically nonfunctional protein product. The remaining variants were compared with published data or were evaluated using in silico analyses. SIFT, MutationTaster, and PolyPhen were used to evaluate missense variants, which were only considered pathogenic if consistently predicted deleterious by all three programs (Table 1). Those variants that did not meet these strict criteria for pathogenicity and are of unknown significance are included in Supplemental Table S1. No germ-line variants were included in the analysis.
Table 1LRRK2 Mutations, Predicted Protein Change, Variant Type, and Domain Affected from TCGA
Processed, normalized, and segmented data from Affymetrix SNP 6.0 (Applied Biosystems, Foster City, CA) arrays were used to generate masked (E.R.P.C.) copy number segment files, and numeric focal-level copy number variation values were obtained using GISTIC2 (Genomic Data Commons Copy Number Variation Analysis Pipeline). Copy number variation with values <−0.3 were categorized as loss and values >0.3 were categorized as gain. Nonsynonymous mutation rates from 988 of 1098 cases computed by the Broad Genome Data Analysis Center using MutSigCV version 0.9 (Broad Institute, Cambridge, MA) were obtained from FireBrowse.
The rate of nonsynonymous gene mutations was determined by the Broad Genome Data Analysis Center using MutSigCV version 0.9.
Whole slide images of diagnostic sections were downloaded from the Genomic Data Commons Data Portal and assessed by at least one breast pathologist on ImageScope. Select clinical data, follow-up times, and vital status were available for 1096 of 1098 patients. Overall survival was defined as date of diagnosis to the date of last clinical follow-up or death. Simple somatic mutations were also downloaded and analyzed from the cBio Cancer Genomics Portal.
Statistical analysis included comparisons between the groups using the χ2 test, U-test (where tests for homoscedasticity failed to demonstrate equal variances), and Mantel-Cox log-rank test.
Curated clinical, mutation, and copy number information was available for 1098 of TCGA project-level breast cancer cases. Because of the large size of LRRK2 (51 exons spanning over 148 Kb) and the stringency of gene-level curation of TCGA bioinformatics pipeline, only 93 cases with complete LRRK2 sequence were available. A total of 28 LRRK2 somatic variants (synonymous, nonsynonymous, and deep intronic) were identified, of which 18 were considered potentially pathogenic in 17 cases, including 11 missense, 3 nonsense, 3 frameshift, and 1 splice-site variant (Table 1). LRRK2 encodes a multidomain protein with an enzymatic core composed of a Ras of complex GTPase domain in tandem with a C-terminal of Roc domain and a kinase domain, which localizes to the cytoplasm and mitochondrial membrane.
A large fraction of the alterations were stop-gain, inactivating mutations (41% of cases). Interestingly, missense variants commonly targeted the Ras of complex, C-terminal of Roc, and kinase domains (Table 1). A single case was characterized by the PD-associated LRRK2 p.G2385R mutation. Cases with wild-type/polymorphic variant LRRK2 sequences and cases with variants with insufficient evidence for pathogenicity are listed in Supplemental Tables S1 and S2.
All patients with tumors harboring LRRK2 mutations were female, had a median age at diagnosis of 61 (range, 41 to 90) years, and most presented with stage II or III disease (82%); these characteristics were similar to the study cohort (Supplemental Table S3).
; interestingly, a granular eosinophilic cytoplasm, consistent with mitochondrial-rich apocrine-like features, was present in 71% of LRRK2-mutated cases (Figure 1A and Table 2). Clinical predictive marker information was available for 15 patients with LRRK2 mutations. Most cases were estrogen receptor–positive disease (67%), including three cases that were additionally positive for human epidermal growth factor receptor 2 (Supplemental Table S3), four cases that were triple negative, and no case that was positive for human epidermal growth factor receptor 2 in the absence of estrogen receptor expression (Supplemental Table S3). Androgen receptor expression information was not available.
Table 2Clinicopathologic Summary of Breast Cancer Patients Harboring an LRRK2 Mutation
TP53 alterations, including stop-gain mutations and copy number losses, were the most common event (65%) (Supplemental Figure S1). Furthermore, LRRK2-mutated carcinomas displayed a high mutation burden compared with the remainder of the breast cancer cohort (mean, 16.6 versus 1.4 nonsilent mutations per megabase, respectively; U-test P = 0.003) (Figure 1B). In fact, the two cases with the highest tumor mutation burdens in TCGA Breast Cancer Project harbored LRRK2 mutations. LRRK2-mutated carcinomas showed similarly high mutation burdens when only cases with complete LRRK2 sequence were considered, despite the small sample size (Supplemental Figure S2). Consistent with a high-risk phenotype, subjects with LRRK2 mutations had a statistically significant inferior overall survival, including 3-year survival rates of 31% and 84% among patients with LRRK2 mutations and the remainder of the TGCA Breast Cancer Project, respectively, with a hazard ratio of 3.42 (95% CI, 1.08–10.80; P = 0.036) (Figure 1C). Similar overall survival trends were observed when LRRK2-sequenced cases alone were examined, albeit this study was underpowered (Supplemental Figure S3).
We uncovered a previously unrecognized role of LRRK2 in breast cancer, whereby tumors harboring a somatic LRRK2 mutation may constitute a distinct clinicopathologic entity with increased tumor mutation burden and high-risk clinical behavior. Immune checkpoint inhibitors have emerged as effective oncologic therapy, leveraging adaptive immune responses, and detecting error-prone DNA repair deficiencies predicts responses to these drugs, presumably due to the production of mutation-associated neoantigens.
Detailed biochemical studies will be necessary to dissect whether LRRK2 alterations are causative of increased mutation burdens or a sensitive marker of LRRK2-independent mutagenesis. Nonetheless, the high fraction of inactivating stop-gain mutations in LRRK2 would suggest a mechanistic relationship, not a bystander effect, and evidence exists supporting a DNA maintenance role for LRRK2.
Assessing for LRRK2 integrity may thus identify a subset of genomically unstable breast tumors with vulnerability to immune checkpoint inhibition.
A recent study reported that PD patients with germ-line LRRK2 p.G2019S mutations were at an increased risk for leukemia and colon cancer, suggesting that the overall baseline cancer risk may be elevated among LRRK2 mutation carriers.
It will be informative to investigate whether LRRK2 mutations occur in other primary sites, supporting a broader tumor suppressor function for LRRK2, and future mechanistic studies may examine a direct role in cancer pathogenesis.
Several PD-associated genes, including LRRK2, have been implicated in cell cycle control and tied to DNA integrity, which are classic tumor suppressor functions frequently targeted for inactivation in cancer.
The somatic LRRK2 variants identified herein were predicted to be loss of function, including a high proportion of stop-gain mutations, albeit biochemical functional characterizations are still necessary to substantiate the pathogenic nature of these variants. Yet, biochemical studies examining the common PD-linked p.G2019S mutation have described a gain of function,
and clinical trials for PD using LRRK2 kinase inhibitors and antisense oligonucleotides are currently underway (clinicaltrials.gov, last accessed July 10, 2020). In light of our results, it may be prudent to assess the impact of LRRK2 dysregulation on carcinogenesis as an adverse effect of LRRK2-directed PD therapy.
One of the limitations of our study is that of the 1098 breast cancer cases in TCGA, only a fraction had available LRRK2 sequence information, limiting the statistical power of these results. To address this, somatic LRRK2 pathogenic variants were identified in the cBio Cancer Genomics Portal in primary and metastatic breast cancer (Supplemental Table S4), showing some overlap with TCGA Breast Cancer Project, including another missense variant targeting leucine 425 and nonsense mutation p.S858∗. The fact that stop-gain mutations and homozygous deletions are reported in this external data set additionally supports our view that LRRK2 inactivation is a breast cancer–associated phenomenon (Supplemental Table S4). Furthermore, as a retrospective study, the findings were limited by the paucity of clinical information available for the evaluation of therapeutic outcomes. Notwithstanding these limitations, the relative frequency of LRRK2 mutations detected is strong evidence for its biomedical and oncologic relevance. On the basis of our collective findings, the inclusion of LRRK2 in commercial somatic sequencing panels may be warranted.
In conclusion, somatic LRRK2 mutations are associated with breast cancer, and these findings shed light onto the mechanistic underpinnings of LRRK2-related neurodegeneration versus neoplastic disease. The identification of LRRK2 mutations in breast cancer may provide an opportunity for risk stratification and new treatment paradigms.
We thank Drs. Michael Lutz, Daniel Laskowitz, and Rex Bentley for thoughtful manuscript review and discussion.
Genomic landscape of LRRK2 carcinomas. ∗Nonsynonymous missense changes are represented by gray boxes, and stop-gain mutations (ie, nonsense, frameshift, and splice site) are black boxes. †Copy number gains are represented by red boxes, and losses are blue boxes. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; Mut., mutations; N, not available; PR, progesterone receptor.