Impact of somatic PIK3CA mutations on clinical outcomes in HER2-positive breast cancer Authors: Nickolas Stabellini, Sonia D de Oliveira, Takae Mizukami, Shawn M Sweeney, Nikolaus Schultz, Ken Kehl, Gregory Riely, Katherine Panageas, Deborah Schrag, Philippe Bedard, Christine Micheel, Xindi Guo, Alberto J Montero Published: 2026 ABSTRACT AbstractBackgroundHER2-positive (HER2+) breast cancer (BC) accounts for 15–20% of all BC. Mutations in the PIK3CA gene are found in 25–30% of HER2 + BC cases and have been implicated in tumor progression and endocrine therapy resistance. Recent evidence suggests that a somatic PIK3CA mutation (PIK3CAm) may serve as a marker of early progression in HER2 + metastatic BC. However, its precise prognostic significance across various lines of therapy and disease stages is not fully understood. We therefore aimed to evaluate the impact of PIK3CAm status on clinical outcomes in HER2 + BC patients.MethodsWe conducted a retrospective cohort study using data from the American Association for Cancer Research (AACR) Project GENIE Biopharma Collaborative (BPC). Overall survival (OS), Distant Recurrence-Free Survival (DRFS), and progression-free survival (PFS) were the primary outcomes. We compared patients with and without PIK3CAm and performed survival analysis using Kaplan-Meier methods and Cox proportional hazards models.ResultsWe identified 212 HER2 + patients (150 early-stage and 62 Stage IV), of which 58 had a PIK3CAm. The median OS for the overall cohort was 125.1 months from treatment. Among early-stage patients, PIK3CAm was associated with a longer DRFS from diagnosis (37.7 months [95% CI 28.6–62.6] vs. 22 months [95% CI 16.8–32.8], p = 0.04). In contrast, among patients with Stage IV disease receiving first-line therapy, PIK3CAm was associated with a shorter median PFS (5.5 [95% CI 1.6–32.2] vs. 14.9 months [95% CI 7.2–22.0], p = 0.04). PIK3CAm was not identified as an independent predictor of OS, DRFS, or PFS in univariable or multivariable regression analyses. Within the PIK3CAm subgroup, not having received hormone therapy was the only factor associated with worse survival on multivariable analysis (aHR = 4.51, 95% CI 1.77–11.5).ConclusionPIK3CAm was not found to be associated with worse OS, however it was associated with a shorter PFS among patients receiving first-line therapy, suggesting PIK3CAm could be explored as a predictive biomarker for identifying HER2 + Stage IV patients at higher risk of disease progression. These findings underscore the need for enhanced surveillance and personalized treatment strategies to manage disease progression in this patient subgroup.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s13058-026-02324-6.Keywords: PIK3CA, Mutation, Survival, Progression, Recurrence, Breast cancer, HER2-positive Introduction Female breast cancer (BC) is the most diagnosed malignancy in the world [1]. Worldwide, in 2022, it was responsible for 2,308,897 cases (11.6% of all cancer cases) and 665,684 related deaths (6.9% of all cancer-related deaths) [1]. In the United States (U.S.), BC was estimated to be responsible for 319,750 new cases and 42,680 related deaths in 2025 [2]. Newly diagnosed BC cases are routinely tested for estrogen (ER) and progesterone (PR) receptor expression, as well as for overexpression of human epidermal growth factor receptor 2 (HER2), which together guide both prognostic evaluation and therapeutic decision-making [3]. Among these, HER2-positive (HER2+) BC accounts for approximately 15–20% of all diagnoses, identifying patients who benefit from HER2-targeted treatments [3, 4]. Historically, HER2 + BC was considered an aggressive subtype, given the greater metastatic potential, but the development of HER2-targeted therapies has dramatically improved survival outcomes [5]. Currently, patients with early-stage HER2 + BC treated with chemotherapy and dual antibody therapy can achieve 3-year invasive-disease-free survival rates above 90%, and more than half of patients with metastatic HER2 + disease are diagnosed de novo BC [5–7]. The phosphatidylinositol-3-kinase (PI3K) pathway, driven by the PIK3CA gene, plays a critical role in tumor initiation, growth, proliferation, and therapy resistance [8]. After the TP53 tumor suppressor gene, the PIK3CA oncogene is the most frequently mutated gene in BC [9–11]. PIK3CA mutations (PIK3CAm) are estimated to occur in about 35% of all BC cases, most commonly in ER-positive and HER2-negative subtypes ] [12]. In HER2 + BC, these mutations are found in approximately 25–30% of cases [13]. PIK3CAm has been reported to be associated with shorter survival times in metastatic breast cancer (mBC), reduced effectiveness of HER2-targeted therapies, and resistance to endocrine therapy [14, 15]. In the recently presented phase 3 Destiny Breast09 trial, the presence of a PIK3CAm was identified as a marker of early progression in HER2 + metastatic BC (mBC) receiving first-line therapy with trastuzumab/pertuzumab monoclonal antibodies [16]. Although PIK3CAm is consistently associated with lower rates of pathological complete response (pCR) in the neoadjuvant setting for HER2 + patients and with a poorer prognosis, its precise prognostic significance across various lines of therapy and disease stages in HER2 + BC is not fully understood. Therefore, we aimed to evaluate the impact of PIK3CAm status on clinical outcomes in HER2 + BC patients. Methods Data source Data was obtained from the American Association for Cancer Research (AACR) Project Genomics Evidence Neoplasia Information Exchange Biopharma Collaborative (GENIE BPC) Breast Cancer Cohort version 1.2 [17–19]. The GENIE BPC is a multi-phase, pre-competitive collaboration between 10 biopharmaceutical companies and select GENIE-participating academic institutions [17]. It focuses on providing detailed clinical annotations for a subset of patients within the GENIE Registry [17]. The data available integrates demographic, diagnosis, genomic, and treatment information with longitudinal, real-world patient outcomes [17]. Data collection follows a structured framework to ensure interoperability and forward compatibility with other data models [17]. Each cohort undergoes a series of rigorous quality control and assurance protocols, which ensures consistency, accuracy, and reliability of the data across multiple institutions before its public release [17]. The biopharma collaborative datasets consist of patient- and sample-level data sourced from the AACR Project GENIE Registry [17]. This is supplemented with additional longitudinal clinical data, which is collected from institutional electronic health records and tumor registries into local REDCap databases [17]. Curators utilize the PRISSMM™ framework for this collection [17, 20, 21]. PRISSMM™ standardizes data collection from Pathology, Radiology, Imaging, Signs and Symptoms, tumor Markers, and Medical oncologist assessments, combining these data with patient demographics and treatments [17, 20, 21]. Study design The study was designed as a retrospective cohort. The BC cohort of the GENIE BPC includes patients diagnosed between 18 and 56 years of age who underwent tumor sequencing between 2013 and 2020. Of those, our study included HER2 + patients who had received at least one line of therapy (Fig. 1). CONSORT diagram of study cohort selection Outcomes The primary outcomes were overall survival (OS), Distant Recurrence-Free Survival (DRFS; calculated for early-stage patients), and progression-free survival (PFS; calculated for those with Stage IV disease). Covariates Patient and clinical characteristics, treatment history, and genomic alterations were included (Supplemental Table 1). Demographics included age at diagnosis, sex, race, and ethnicity. Tumor Characteristics encompassed cancer sequence, tumor laterality, stage at diagnosis, BC histology, ER status, PR status, HER2 status, and BC subtype. Treatment-Related Factors comprised neoadjuvant therapy before pathologic staging, number of regimens, details of the first regimen (e.g., number of drugs, clinical trial participation, duration), and the receipt of specific therapy types (chemotherapy, hormone therapy, targeted/immunotherapy; defined by drug names according to Supplemental Table 2). Distant Metastasis included the presence and site of distant metastasis at diagnosis. Clinical Procedures and Markers involved the number of imaging reports, types of scans received (CT, MRI, PET, Bone Scan, Mammogram), number of pathology reports, number of medical oncologist visits, and tumor marker results (CA 15 − 3, CA 27.29). Genomic Alterations included the alteration status of key genes (e.g., TP53, PIK3CA, ERBB2) and pathways (e.g., p53, PI3K, cell cycle, RTK/RAS). Sequencing Data provided details about the cancer panel tests (e.g., year of sequencing, time from diagnosis to report, specimen histological subtype). Statistical analysis Descriptive analysis summarized continuous variables as mean and standard deviation (SD) if normally distributed, or as median and interquartile range (IQR; 25th and 75th percentiles) if non-normally distributed. Data distribution was assessed using the Shapiro-Wilk test and histograms. Categorical variables were summarized as frequencies and percentages. Comparative statistics, comparing patients with PIK3CAm versus those without, were performed using Fisher’s exact test for categorical variables. For continuous covariates, independent samples t-tests were used for normally distributed variables, and the non-parametric Kruskal-Wallis test was used for non-normally distributed variables. Survival analyses were performed using Kaplan-Meier to estimate OS, DRFS, and PFS. The log-rank test was used to compare survival distributions overall and across subgroups, including PIK3CA mutation status, ER status (positive vs. negative), and TNM stage (early-stage [TNM I–III] vs. Stage IV), with TNM analyses also performed stratified by line of therapy (first- vs. second-line). Univariable Cox proportional hazards (COX PH) models were used to assess the association between individual covariates and outcomes, and results were reported as hazard ratios (HR), 95% confidence intervals (CI), and p-values. Multivariable COX PH were constructed using a LASSO (Least Absolute Shrinkage and Selection Operator) regression approach for variable selection. Variables were first filtered based on a minimum alteration frequency (for genomic data) and then for near-zero variance. Highly correlated variables (Pearson correlation coefficient > 0.70) were also identified and handled during the LASSO process to ensure model stability. A 10-fold cross-validation was performed to identify the optimal penalty parameter (λ). The λ min (lambda.min) value, which corresponds to the λ that gives the minimum mean cross-validated error, was chosen for selecting the final set of covariates. The selected variables from the LASSO model were then included in the final multivariable COX PH models. All models were performed after checking if assumptions were met. Cox models for OS were performed in the overall cohort and stratified by ER status, TNM stage, and line of therapy. For DRFS and PFS, stratified analyses by ER status and line of therapy were not performed. No interaction analyses were conducted. All statistical analyses were performed using R (version 4.5.0) [22]. A p-value of < 0.05 was considered statistically significant. Results We included 212 HER2 + patients (Table 1; Supplemental Tables 3–6), with a median age at diagnosis of 43 years (IQR 37–48). Most were female (99.1%), White (77%) and non-Hispanic (95%), and presented with Stage II disease (12%). 98% of patients received cytotoxic chemotherapy, 92% received HER2 therapy, 67% endocrine therapy, 14% CDK4/6 inhibitors, 9% PIK3/mTOR inhibitors, and 9% received immunotherapy. Table 1Overall patient characteristics of HER2 + patient cohort and by PIK3CA Mutation statusCharacteristicOverallPIK3CA not alteredPIK3CA alteredp-value2N = 2121N = 1541N = 581Age at diagnosis43 (37, 48)42 (37, 48)43 (37, 46)0.8Race–––0.8White163 (77%)117 (76%)46 (79%)–Black16 (7.5%)12 (7.8%)4 (6.9%)–Other12 (5.7%)8 (5.2%)4 (6.9%)–Unknown/Other21 (9.9%)17 (11%)4 (6.9%)–Ethnicity–––0.9Hispanic7 (3.3%)6 (3.9%)1 (1.7%)–Non-Hispanic201 (95%)145 (94%)56 (97%)–Unknown4 (1.9%)3 (1.9%)1 (1.7%)–Stage at diagnosis–––0.5Stage I-III150 (71%)107 (69%)43 (74%)–Stage IV62 (29%)47 (31%)15 (26%)–Histology–––0.7Invasive ductal carcinoma204 (96%)148 (96%)56 (97%)–Invasive lobular carcinoma6 (2.8%)5 (3.2%)1 (1.7%)–Other histology2 (0.9%)1 (0.6%)1 (1.7%)–ER status––– 0.007 Negative80 (38%)67 (44%)13 (22%)–Positive132 (62%)87 (56%)45 (78%)–PR status–––0.06Negative90 (42%)72 (47%)18 (31%)–Positive121 (57%)81 (53%)40 (69%)–Unknown/Not tested1 (0.5%)1 (0.6%)0 (0%)–PIK3CA mutation type––– < 0.001 Kinase domain (H1047)24 (11%)0 (0%)24 (41%)–Helical domain (E545)7 (3.3%)0 (0%)7 (12%)–Other PIK3CA mutation26 (12%)0 (0%)26 (45%)–Received chemotherapy207 (98%)152 (99%)55 (95%)0.13Received hormone therapy142 (67%)97 (63%)45 (78%) 0.04 Received immunotherapy19 (9.0%)14 (9.1%)5 (8.6%)> 0.9Received HER2 therapy194 (92%)145 (94%)49 (84%) 0.04 Received CDK4/6 inhibitors29 (14%)17 (11%)12 (21%)0.07Received PI3K/mTOR inhibitors19 (9.0%)11 (7.1%)8 (14%)0.21Median (Q1, Q3); n (%)The bold p-values indicate statistical significance (p < 0.05) Overall patient characteristics of HER2 + patient cohort and by PIK3CA Mutation status 1Median (Q1, Q3); n (%) The bold p-values indicate statistical significance (p < 0.05) Of the 212 patients included in our study, 58 had a PIK3CAm, of which 24 (41%) were kinase domain mutations (Table 1). Comparing those with PIK3CAm versus those without (Table 1; Supplemental Tables 3–6), patients with PIK3CAm had higher rates of ER-positive disease (78% vs. 56%, p = 0.007), and higher rates of receiving endocrine therapy (78% vs. 63%, p = 0.04). Conversely, the presence of a PIK3CAm was associated with lower rates of HER2 therapy (84% vs. 94%, p = 0.04). The time from initial diagnosis to distant metastasis in early-stage patients was significantly longer for those with a PIK3CAm (median: 38 months [IQR: 23–73] vs. 22 months [IQR: 10–43], p = 0.01; Supplemental Table 3). Furthermore, as expected, patients with PIK3CAm had higher rates of altered PI3K pathway (100% vs. 21%, p < 0.001). Interestingly, they also had significantly lower rates of Hippo pathway alteration (0% vs. 11%, p = 0.01). Impact of PIK3CAm on survival outcomes The median OS from initial BC diagnosis was 124.4 months (95% CI 98.4-157.2; Table 2). In univariable COX PH analysis, PIK3CAm was not significantly associated with worse OS outcomes (HR = 1.06, 95% CI 0.68–1.66; Table 3). LASSO did not select PIK3CAm for the set of covariates in multivariable COX PH analysis (Table 3). Similar results were found in subgroup analyses by stage and by ER-status (Supplemental Tables 7–10). Table 2Survival Summary by HER2 + subgroup and endpointMedian Survival2-Year Survival5-Year Survival10-Year SurvivalStatus N Eventsp-valueMedian survival (Months)95% CI lower95% CI upper2-Year prob.95% CI lower95% CI upper5-year prob.95% CI lower95% CI upper10-year prob.95% CI lower95% CI upper HER2+ OS from diagnosisPIK3CA altered58290.80712698.8NA96.50%0.918169.90%0.5860.83551.90%0.390.691PIK3CA not altered15456120.987.9157.297.30%0.9470.99981.40%0.7460.88851.40%0.4130.641Overall21285-124.498.4157.297.10%0.9480.99477.90%0.7190.84451.60%0.4330.615OS from treatmentPIK3CA altered57280.666125.198.3NA96.40%0.917171.50%0.6020.84952.20%0.390.7PIK3CA Not altered15255131.287.7157.795.90%0.9270.99278.50%0.7120.86654.80%0.4480.67Overall20983-125.198.3155.396.00%0.9340.98876.50%0.7030.83353.80%0.4550.637 HER2 + ER negative OS from diagnosisPIK3CA altered1380.23644.937.6NA100.00%1136.70%0.1680.79818.30%0.0370.898PIK3CA Not altered672986.465NA93.80%0.880.99970.90%0.5940.84535.80%0.2160.593Overall8037-85.864.2140.194.80%0.8990.99964.80%0.5390.7832.80%0.2020.533OS from treatmentPIK3CA altered1270.2474434.8NA100.00%1140.40%0.190.85920.20%0.0420.979PIK3CA not altered662885.266.1NA90.40%0.8340.9871.80%0.6030.85545.40%0.3120.66overall7835-85.265.414291.90%0.8580.98366.90%0.5590.79941.30%0.2830.603 HER2 + ER positive OS from diagnosisPIK3CA altered45210.597152.9103.3NA95.60%0.897178.70%0.6720.92259.80%0.4550.785PIK3CA not altered8727153.198.4NA100.00%1189.30%0.8190.97362.50%0.4990.782Overall13248-153.1120.9168.998.50%0.964185.30%0.7880.92361.10%0.5110.729OS from treatmentPIK3CA altered45210.565151.3102.1NA95.60%0.897178.70%0.6720.92258.90%0.4440.782PIK3CA not altered8627151.397.4NA100.00%1183.50%0.7450.93661.80%0.4910.779Overall13148-151.3120.3167.198.40%0.963181.90%0.7470.89760.40%0.5040.725 HER2 + Early stage (I-III) OS from diagnosisPIK3CA altered43220.907152.998.8NA100.00%1174.90%0.6250.89755.50%0.410.751PIK3CA not altered10741134.886.4168.496.10%0.9250.99978.80%0.7040.88353.70%0.4240.679Overall15063-134.898.416397.30%0.9470.99977.60%0.7050.85454.10%0.4490.653OS from treatmentPIK3CA altered42210.955151.398.3NA100.00%1177.10%0.650.91556.10%0.4120.764PIK3CA not altered10641131.287.3162.896.10%0.9240.99973.80%0.6450.84555.00%0.4370.692Overall14862-131.298.3160.697.20%0.9460.99975.10%0.6760.83455.20%0.4580.665DRFS from diagnosisPIK3CA altered3434 0.048 37.728.662.673.50%0.6010.935.30%0.2240.5562.90%0.0040.203PIK3CA not altered81812216.832.848.10%0.3840.60418.50%0.1170.2923.70%0.0120.112Overall115115-26.321.835.555.70%0.4730.65523.50%0.1690.3273.50%0.0130.091DRFS from treatmentPIK3CA altered32320.11136.927.761.771.90%0.5790.89334.40%0.2130.5553.10%0.0050.215PIK3CA not altered727222.419.233.545.80%0.3570.58919.40%0.1220.3114.20%0.0140.126Overall104104-26.921.935.653.80%0.4510.64324.00%0.1710.3383.80%0.0150.101 HER2 + Early stage − 1st Line OS from diagnosisPIK3CA altered40190.896163101.9NA100.00%1175.80%0.6320.9157.60%0.4250.78PIK3CA not altered10440134.887.9168.496.00%0.9230.99978.50%0.6990.88154.60%0.4320.689Overall14459-134.898.8163.797.20%0.9440.99977.70%0.7050.85755.40%0.460.668OS from treatmentPIK3CA altered40NANANANANANANANANANANANANANAPIK3CA not altered104NANANANANANANANANANANANANANAOverall144NA-NANANANANANANANANANANANADRFS from diagnosisPIK3CA altered33330.07736.627.162.672.70%0.590.89633.30%0.2060.543.00%0.0040.209PIK3CA not altered808022.916.832.848.70%0.3890.6118.70%0.1190.2963.70%0.0120.114Overall113113-26.321.835.555.80%0.4730.65723.00%0.1640.3223.50%0.0140.093DRFS from treatmentPIK3CA altered31310.1635.42761.771.00%0.5670.88932.30%0.1940.5373.20%0.0050.222PIK3CA not altered717122.419.333.546.50%0.3620.59719.70%0.1230.3154.20%0.0140.128Overall102102-26.922.335.453.90%0.4510.64523.50%0.1660.3343.90%0.0150.102 HER2 + Early stage − 2nd line OS from diagnosisPIK3CA altered40200.867152.9101.9NA100.00%1176.40%0.640.91258.90%0.440.787PIK3CA not altered9740134.886.4163.795.80%0.9180.99977.50%0.6870.87453.50%0.4210.679Overall13760-134.898.816397.00%0.9420.99977.20%0.6980.85255.20%0.4580.665OS from TreatmentPIK3CA altered40NANANANANANANANANANANANANANAPIK3CA not altered97NANANANANANANANANANANANANANAOverall137NA-NANANANANANANANANANANANADRFS from diagnosisPIK3CA altered33330.05838.928.672.972.70%0.590.89636.40%0.2320.5713.00%0.0040.209PIK3CA not altered78782416.833.350.00%0.40.62419.20%0.1220.3033.80%0.0130.117Overall111111-2823.436.656.80%0.4820.66824.30%0.1750.3383.60%0.0140.094DRFS from treatmentPIK3CA altered31310.13338.527.770.571.00%0.5670.88935.50%0.2210.573.20%0.0050.222PIK3CA not altered696922.719.333.747.80%0.3740.61220.30%0.1270.3244.30%0.0140.131Overall100100-28.422.438.555.00%0.4610.65725.00%0.1780.3514.00%0.0150.104 HER2 + Stage IV OS from diagnosisPIK3CA altered1570.595114.853.1NA85.70%0.692155.60%0.3430.89941.70%0.1980.876PIK3CA not altered4715117.381.3NA100.00%1187.20%0.7720.98641.60%0.2270.764Overall6222-114.881.3NA96.60%0.922178.60%0.6790.9142.10%0.2610.677OS from treatmentPIK3CA altered1570.369112.151.1NA85.70%0.692155.60%0.3430.89941.70%0.1980.876PIK3CA not altered4614120.380.8NA95.30%0.893189.30%0.796152.20%0.3380.805Overall6121-112.180.8NA93.00%0.8670.99979.90%0.6930.92248.30%0.3230.723PFS from diagnosisPIK3CA altered15150.2297.11.632.220.00%0.0730.556.70%0.010.4430.00%NANAPIK3CA not altered4742157.223.231.90%0.210.48512.00%0.0490.2920.00%NANAOverall6257-11.67.118.529.00%0.1970.42810.30%0.0450.2350.00%NANAPFS from treatmentPIK3CA altered1570.90428.76.4NA50.60%0.2650.96725.30%0.05510.00%NANAPIK3CA not altered442014.27.9NA34.20%0.1890.6219.00%0.0660.54819.00%0.0660.548Overall5927-14.26.955.636.80%0.2250.60220.70%0.0880.4840.00%NANA HER2 + Stage IV − 1st line OS from diagnosisPIK3CA altered1470.214114.837.6NA84.60%0.671151.90%0.3030.89134.60%0.1320.909PIK3CA not altered4413120.981.3NA100.00%1188.80%0.788150.60%0.320.799Overall5820-114.881.3NA96.40%0.916178.90%0.6780.91744.30%0.2760.71OS from treatmentPIK3CA altered14NANANANANANANANANANANANANANAPIK3CA not altered44NANANANANANANANANANANANANANAOverall58NA-NANANANANANANANANANANANAPFS from diagnosisPIK3CA altered1414 0.042 5.51.632.214.30%0.040.5150.00%NANA0.00%NANAPIK3CA not altered443914.97.22229.50%0.1870.46612.80%0.0530.3110.00%NANAOverall5853-8.97.118.225.90%0.1670.48.90%0.0340.2320.00%NANAPFS from treatmentPIK3CA altered14NANANANANANANANANANANANANANAPIK3CA not altered44NANANANANANANANANANANANANANAOverall58NA-NANANANANANANANANANANANA HER2 + Stage IV − 2nd line OS from diagnosisPIK3CA altered1370.254114.837.6NA84.60%0.671151.90%0.3030.89134.60%0.1320.909PIK3CA not altered3813120.981.3NA100.00%1187.80%0.772150.00%0.3160.792Overall5120-114.881.3NA96.00%0.908177.60%0.6610.91143.50%0.2710.701OS from treatmentPIK3CA altered13NANANANANANANANANANANANANANAPIK3CA not altered38NANANANANANANANANANANANANANAOverall51NA-NANANANANANANANANANANANAPFS from diagnosisPIK3CA altered13130.1473.91NA15.40%0.0430.550.00%NANA0.00%NANAPIK3CA not altered38361232226.30%0.1550.4487.90%0.0230.2690.00%NANAOverall5149-7.43.216.323.50%0.1430.3865.50%0.0150.1960.00%NANAPFS from treatmentPIK3CA altered13NANANANANANANANANANANANANANAPIK3CA not altered38NANANANANANANANANANANANANANAOverall51NA-NANANANANANANANANANANANAp-values calculated using the log-rank test comparing PIK3CA Altered vs. Not Altered groups. Survival probabilities estimated using Kaplan-Meier method.The bold p-values indicate statistical significance (p < 0.05) Survival Summary by HER2 + subgroup and endpoint p-values calculated using the log-rank test comparing PIK3CA Altered vs. Not Altered groups. Survival probabilities estimated using Kaplan-Meier method. The bold p-values indicate statistical significance (p < 0.05) Table 3Univariable and multivariable Cox Proportional Hazard Models for Overall Survival after diagnosis for HER2 + patientsCharacteristicUnivariableMultivariable N HR95% CIp-valueaHR95% CIp-valueAge at diagnosis21211.01, 1.08 0.015 1.020.99, 1.060.2Race212––––––White–——–——–Black–1.40.68, 2.750.41.220.59, 2.540.6Asian–0.70.21, 2.150.50.840.23, 3.010.8Other/Unknown–0.70.26, 1.990.50.460.16, 1.340.2Ethnicity212––––––Non-Hispanic––Hispanic–0.70.18, 2.980.7–––Unknown–0.70.10, 5.180.7–––Tumor Laterality212––––––Left––––Other/Unknown–1.40.79, 2.490.31.640.87, 3.080.12Right–0.80.47, 1.250.30.770.46, 1.290.3Histology212——————Invasive ductal carcinoma———————Invasive lobular carcinoma—2.50.77, 7.840.132.090.63, 6.880.2Other histology—4.51.10, 18.7 0.036 4.290.89, 20.70.069ER Status212——————Positive———Negative—2.21.44, 3.49 < 0.001 BorderlineUnknown/Not testedPR Status212Positive————Negative2.21.45, 3.46 < 0.001 1.70.69, 4.190.2Unknown/Not tested00.00, Inf> 0.900.00, Inf> 0.9Stage at diagnosis212Early stage (I-III)——Stage IV1.20.72, 1.930.5Unknown/Other stageBreast cancer subtype212HER2+, HR-————HER2+, HR+0.40.26, 0.64 < 0.001 1.430.50, 4.080.5Received chemotherapy212Yes——No2.20.55, 9.230.3Received hormone therapy212Yes————No2.81.76, 4.46 < 0.001 3.231.55, 6.71 0.002 Received immunotherapy212Yes————No0.40.23, 0.76 0.004 0.550.29, 1.030.062Received HER2 therapy212Yes————No0.60.27, 1.130.10.310.12, 0.76 0.011 Received CDK4/6 inhibitors212Yes——No1.30.77, 2.320.3Received PI3K inhibitors212Yes——No1.30.67, 2.410.5Received other therapy212Yes——No2.30.73, 7.400.2TP53 alteration212Altered——Not altered0.70.41, 1.030.068PIK3CA alteration2121.10.68, 1.660.8ERBB2 alteration212Altered————Not altered1.30.80, 1.990.32.211.25, 3.91 0.006 p53 pathway212Altered——Not altered0.70.44, 1.200.2PI3K pathway212Altered——Not altered1.30.83, 1.980.3Cell cycle pathway212Altered——Not altered0.90.55, 1.390.6CI, Confidence interval; HR, Hazard ratio; aHR, adjusted Hazard RatioThe bold p-values indicate statistical significance (p < 0.05) Univariable and multivariable Cox Proportional Hazard Models for Overall Survival after diagnosis for HER2 + patients CI, Confidence interval; HR, Hazard ratio; aHR, adjusted Hazard Ratio The bold p-values indicate statistical significance (p < 0.05) The median OS from treatment (Table 2) was 125.1 months (95% CI, 98.3–155.3) for the overall cohort: 131.2 months (95% CI, 98.3–160.6) for early-stage patients, and 112.1 months (95% CI, 80.8–NA) for those with stage IV disease. Univariable COX PH analysis showed that PIK3CAm was not associated with OS (HR = 1.11, 95% CI 0.70–1.75; Table 4). LASSO did not select PIK3CAm for the set of covariates in multivariable COX PH analysis (Table 4). Similar results were found in subgroup analyses by stage and by ER-status (Supplemental Tables 11–14). Table 4Univariable and multivariable Cox Proportional Hazard Models for Overall Survival after treatment for HER2 + patientsCharacteristicUnivariableMultivariable N HR95% CIp-valueaHR95% CIp-valueAge at Diagnosis2091.041.00, 1.07 0.026 1.020.99, 1.060.2Race209White——Black1.40.68, 2.770.4Asian0.70.22, 2.300.6Other/Unknown0.90.31, 2.350.8Ethnicity209Non-Hispanic——Hispanic0.80.19, 3.160.7Unknown1.10.16, 8.23> 0.9Tumor Laterality209Left————Other/Unknown1.10.58, 2.030.81.240.63, 2.430.5Right0.80.46, 1.230.30.750.45, 1.240.3Histology209Invasive ductal carcinoma————Invasive lobular carcinoma2.50.79, 8.080.121.970.59, 6.510.3Other histology4.81.17, 19.9 0.03 5.291.12, 25.0 0.035 ER Status209Positive——Negative1.91.20, 2.91 0.006 BorderlineUnknown/Not testedPR Status209Positive————Negative21.32, 3.15 0.001 1.070.59, 1.950.8Unknown/Not tested00.00, Inf> 0.900.00, Inf> 0.9Stage at Diagnosis209Early stage (I-III)——Stage IV10.60, 1.67> 0.9Unknown/Other stageBreast cancer subtype209HER2+, HR-——HER2+, HR+0.50.30, 0.74 0.001 Received chemotherapy209Yes——No00.00, Inf> 0.9Received hormone therapy209Yes————No2.81.73, 4.50 < 0.001 2.551.33, 4.86 0.005 Received immunotherapy209Yes————No0.40.24, 0.79 0.006 0.480.25, 0.91 0.025 Received HER2 therapy209Yes————No0.50.21, 1.050.0640.40.16, 0.99 0.048 Received CDK4/6 inhibitors209Yes————No1.50.84, 2.620.21.730.85, 3.530.13Received PI3K inhibitors209Yes——No1.50.74, 2.830.3Received other therapy209Yes——No2.20.68, 6.880.2TP53 alteration209Altered——Not altered0.60.40, 1.010.056PIK3CA alteration2091.10.70, 1.750.7ERBB2 alteration209Altered————Not altered1.20.75, 1.920.42.271.23, 4.16 0.008 p53 pathway209Altered——Not altered0.80.47, 1.300.3PI3K pathway209Altered——Not altered1.20.77, 1.860.4Cell cycle pathway209Altered——Not altered0.90.58, 1.500.8CI, Confidence interval; HR = Hazard ratio; aHR = adjusted hazard ratioThe bold p-values indicate statistical significance (p < 0.05) Univariable and multivariable Cox Proportional Hazard Models for Overall Survival after treatment for HER2 + patients CI, Confidence interval; HR = Hazard ratio; aHR = adjusted hazard ratio The bold p-values indicate statistical significance (p < 0.05) The median DRFS and PFS from initial BC diagnosis (Table 2) were 26.3 months (95% CI, 21.8–35.5) and 11.6 months (95% CI, 7.1–18.5), respectively. Univariable COX PH analysis showed that PIK3CAm was not associated with a higher risk of distant recurrence (HR = 0.67, 95% CI 0.44–1.01) or progression (HR = 1.44, 95% CI 0.79–2.63); Supplemental Tables 15 and 16). PIK3CAm was not selected by LASSO for the multivariable analysis. The median DRFS and PFS from the time of first treatment (Table 2) were 26.9 months (95% CI 21.9–35.6) and 14.2 months (95% CI 6.9–55.6), respectively. Univariable COX PH analysis showed that PIK3CAm was not associated with distant recurrence (HR = 0.7, 95% CI 0.44–1.01; Table 5) or progression (HR = 0.95, 95% CI 0.38–2.36; Table 6). LASSO did not select PIK3CAm for multivariable analyses (Tables 5 and 6). Table 5Univariable and multivariable Cox Proportional Hazard Models for Distant Recurrence-Free Survival after treatment for Early Stage HER2 + patientsCharacteristicUnivariableMultivariable N HR95% CIp-valueaHR95% CIp-valueAge at Diagnosis1151.11.02, 1.08 0.002 1.031.00, 1.060.053Race115White——Black1.20.65, 2.310.5Asian0.50.21, 1.320.2Other/Unknown1.60.91, 2.890.1Ethnicity115Non-Hispanic——Hispanic0.90.29, 2.930.9Unknown1.30.42, 4.170.6Tumor Laterality115Left——Other/Unknown1.40.81, 2.280.3Right10.63, 1.430.8Histology115Invasive ductal carcinoma——Invasive lobular carcinoma2.80.68, 11.60.2Other histology1.40.34, 5.650.7ER Status115Positive——Negative2.21.48, 3.34 < 0.001 BorderlineUnknown/Not TestedPR Status115Positive——Negative1.91.28, 2.78 0.001 Unknown/Not tested1.60.22, 11.80.6Breast cancer subtype115HER2+, HR-——HER2+, HR+0.50.30, 0.68 < 0.001 Received hormone therapy115Yes————No2.91.84, 4.43 < 0.001 2.11.31, 3.37 0.002 Received immunotherapy115Yes——No0.60.38, 1.070.086Received HER2 therapy115Yes————No0.40.18, 0.68 0.002 0.50.25, 1.030.059Received CDK4/6 inhibitors115Yes————No2.11.28, 3.28 0.003 1.430.85, 2.400.2Received PI3K inhibitors115Yes——No1.30.72, 2.290.4Received other therapy115Yes——No1.40.86, 2.380.2TP53 alteration115Altered——Not altered0.60.42, 0.92 0.017 PIK3CA Alteration1150.70.44, 1.000.052ERBB2 Alteration115Altered——Not altered0.70.46, 1.040.074p53 pathway115Altered——Not altered0.60.36, 0.87 0.01 PI3K pathway115Altered——Not altered1.40.96, 2.020.079Cell cycle pathway115Altered——Not altered1.10.76, 1.660.6CI, Confidence interval; HR, Hazard ratio; aHR, adjusted Hazard RatioThe bold p-values indicate statistical significance (p < 0.05) Univariable and multivariable Cox Proportional Hazard Models for Distant Recurrence-Free Survival after treatment for Early Stage HER2 + patients CI, Confidence interval; HR, Hazard ratio; aHR, adjusted Hazard Ratio The bold p-values indicate statistical significance (p < 0.05) Table 6Univariable and multivariable Cox Proportional Hazard Models for Progression Free Survival after treatment for Stage IV HER2 + patientsCharacteristicUnivariableMultivariable N HR95% CIp-valueaHR95% CIp-valueAge at Diagnosis5910.97, 1.090.3Race59White——Black0.50.07, 3.970.5Asian2.90.66, 13.10.2Other/Unknown2.10.26, 16.30.5Ethnicity59Non-Hispanic————Hispanic1.10.14, 8.10> 0.91.130.15, 8.54> 0.9Unknown563.47, 890 0.005 71.64.46, 1,150 0.003 Tumor Laterality59Left——Other/Unknown0.70.23, 2.210.6Right0.80.35, 1.930.6Histology59Invasive ductal carcinoma——Invasive lobular carcinoma1.30.30, 5.490.7Other histologyER Status59Positive——Negative10.45, 2.20> 0.9BorderlineUnknown/Not testedPR Status59Positive——Negative1.30.59, 2.760.5Unknown/Not testedBreast cancer subtype59HER2+, HR-——HER2+, HR+0.90.40, 1.960.8Received chemotherapy59Yes——No00.00, Inf> 0.9Received hormone therapy59Yes——No10.46, 2.19> 0.9Received immunotherapy59Yes————No0.10.02, 0.43 0.002 0.080.02, 0.38 0.001 Received HER2 therapy59Yes————No00.00, Inf> 0.900.00, Inf> 0.9Received CDK4/6 inhibitors59Yes——No10.22, 4.15> 0.9Received PI3K inhibitors59Yes——No2.60.34, 19.10.4Received other therapy59Yes——No0.50.13, 1.640.2TP53 alteration59Altered——Not altered0.80.33, 1.960.6PIK3CA Alteration5910.38, 2.36> 0.9ERBB2 Alteration59Altered——Not altered1.40.47, 4.090.5p53 pathway59Altered——Not altered10.38, 2.39> 0.9PI3K pathway59Altered——Not altered1.70.70, 3.970.3Cell cycle pathway59Altered——Not altered0.90.32, 2.300.8CI, Confidence interval; HR, Hazard ratio; aHR, adjusted Hazard RatioThe bold p-values indicate statistical significance (p < 0.05) Univariable and multivariable Cox Proportional Hazard Models for Progression Free Survival after treatment for Stage IV HER2 + patients CI, Confidence interval; HR, Hazard ratio; aHR, adjusted Hazard Ratio The bold p-values indicate statistical significance (p < 0.05) Survival outcomes on patients with PIK3CAm The median OS from initial diagnosis was 126 months (95% CI 98.8-NA; Table 2) for those with PIK3CAm. For these patients, none of the covariates selected for the multivariable model were associated with worse OS (Supplemental Table 17). On subgroup analysis by 1st or 2nd line of therapy (Supplemental Table 17), not having received hormone therapy was associated with worse survival (aHR = 4.65, 95% CI 1.78–12.1 for 1st line and aHR = 4.97, 95% CI 1.91–12.9 for 2nd line). There were not enough patients to stratify analysis by stage and ER-status. The median OS from treatment was 125.1 months (95% CI 98.3-NA; Table 2) for those with a PIK3CAm. For these patients, as expected, not having received HER2-therapy was associated with worse survival (aHR = 4.51, 95% CI 1.77–11.5; Supplemental Table 18). On subgroup analysis by 1st or 2nd line of therapy (Supplemental Table 18), not having received hormone therapy was again associated with worse survival (aHR = 4.67, 95% CI 1.79–12.2 for 1st line and aHR = 5.00, 95% CI 1.89–13.3 for 2nd line). Patient numbers were insufficient for stratification by stage and ER-status. Among patients with early-stage disease, PIK3CAm was associated with longer DRFS from diagnosis (37.7 months [95% CI 28.6–62.6] for PIK3CAm vs. 22.0 months [95% CI 16.8–32.8] for non-PIK3CAm; p = 0.04). In contrast, among patients with Stage IV disease receiving first-line therapy, PIK3CAm was associated with shorter median PFS (5.5 months [95% CI 1.6–32.2] for PIK3CAm vs. 14.9 months [95% CI 7.2–22.0] for non-PIK3CAm; p = 0.04). The median DRFS from treatment for those with PIK3CAm (Table 2) was 36.9 months (95% CI 27.7–61.7). There were not enough patients and/or follow-up to obtain PFS from treatment on patients with PIK3CAm. Discussion In this retrospective cohort study that utilized real-world data from the GENIE BPC, we evaluated the clinical impact of PIK3CAm in 212 HER2 + BC patients receiving first- and second-line therapies. Our primary findings revealed that PIK3CAm was not significantly associated with worse OS, PFS, or distant-recurrence risk, according to COX PH regressions. However, Stage IV HER2 + patients with PIK3Cam receiving first-line therapy had a statistically lower PFS with first line therapy compared to wild-type PIK3CA (median PFS from diagnosis: 5.5 months for PIK3CAm vs. 14.9 months for wild-type, p = 0.04). In addition to the main findings, patients with PIK3CAm in this cohort had higher rates of ER-positive disease (78% vs. 56%, p = 0.007) and lower rates of receiving HER2-targeted therapies (84% vs. 94%, p = 0.04) compared to those with wild-type PIK3CA. Interestingly, early-stage patients with PIK3CAm also showed a longer median time from diagnosis to distant metastasis (38 months vs. 22 months, p = 0.01) compared to those without. Prior studies have reported associations between PIK3CAm and worse outcomes in HER2 + BC [15, 23]. In early-stage HER2 + BC, the NeoALTTO, GeparSixto, and NeoSphere trials reported up to a 20% decrease in pCR rates in PIK3CAm versus wild-type patients after neoadjuvant therapy with HER2-targeted agents, while the APHINITY trial showed poorer clinical outcomes after adjuvant trastuzumab and pertuzumab in patients with PIK3 alterations [24–27]. A meta-analysis reported that PIK3CAm was associated with lower probability of pCR (OR 0.23, 95% CI 0.19–0.27) in HER2 + BC treated with anti-HER2 therapy in the neoadjuvant setting, independent of single- or dual-agent anti-HER2 therapy [23]. The same meta-analysis found no impact of PIK3CAm on disease free survival in the neoadjuvant (HR 1.19, 95% CI 0.85–1.63) or adjuvant (HR 0.84, 95% CI 0.66–1.08) settings, or on invasive disease-free survival (HR 1.24, 95% CI 0.93–1.65), although only a few studies contributed to these outcomes [23]. In HER2 + mBC patients, PIK3CAm was also not associated with worse OS (HR 1.29, 95% CI 0.79–2.11), consistent with our findings (23). Regarding PFS, PIK3CAm was associated with a shorter PFS among patients treated with docetaxel, trastuzumab, and pertuzumab or capecitabine plus lapatinib in CLEOPATRA and EMILIA, respectively [28–30]. Conversely, ExteNET and KATHERINE showed no impact of PIK3CAm in early-stage HER2 + BC receiving adjuvant neratinib or T-DM1, and EMILIA and TH3RESA reported no impact of PIK3CAm on PFS in HER2 + mBC treated with T-DM1 [31–33]. The observed discordance between significantly shorter PFS among patients receiving first-line therapy and the lack of OS differences in Stage IV PIK3CAm in our study may reflect the fact that patients who progress on one therapy can subsequently benefit from other effective therapies, thereby extending their overall lifespan [34]. The continuously evolving therapeutic landscape for HER2 + BC, including advanced antibody-drug conjugates and tyrosine kinase inhibitors, may mitigate the early negative impact of PIK3CA mutations on long-term survival in this patient population, allowing for effective disease management post-progression [5, 35]. In addition, despite PIK3CA being a known oncogenic driver, its limited impact on OS in our cohort, especially when stratified by TNM stage, underscores the dominant prognostic role of TNM staging, which can supersede the influence of individual molecular alterations like PIK3CAm, particularly in a context of effective sequential treatment options [36]. While PIK3CAm may signal a more aggressive tumor biology that may lead to earlier progression, the ability to manage the disease through subsequent lines of therapy may attenuate a potential negative effect on OS in HER2 + BC. Within the PIK3CAm cohort analyzed, the association of not having received hormone therapy with worse OS highlights the critical role of hormone receptor signaling in influencing long-term survival, even in HER2 + disease, and underscores the importance of endocrine therapy where applicable [37–39]. In contrast, our finding that other concurrent mutations were not associated with worse survival in patients with PIK3CAm contrasts with results from a large analysis of over 5,600 mBC patients, in which 42% of those with ERBB2 mutations also had a concurrent PIK3CAm, and the presence of an ERBB2 mutation in this population was associated with a better clinical response to anti-HER2 targeted therapies [40]. Additionally, the finding that patients with PIK3CAm received lower rates of HER2-targeted therapy aligns with evidence suggesting that PIK3CAm may confer resistance to agents such as trastuzumab and lapatinib [15, 23, 41, 42]. However, the biological mechanisms underlying increased resistance to HER2 therapy in the presence of PIK3CAm remain poorly understood, and current evidence on this relationship is inconclusive [15, 23, 42]. Conversely, our observation that PIK3CAm patients exhibited higher rates of ER-positive disease is consistent with prior literature, which consistently reports PIK3CA mutations as more prevalent in ER-positive tumors [8, 11, 43, 44]. One interesting finding was the lower rates of Hippo pathway alterations in patients with PIK3CAm. The Hippo pathway, one of the most frequently dysregulated signaling pathways in human cancer, is a kinase cascade that regulates cell proliferation, differentiation, and survival [45]. In cancer, it plays a critical role in controlling contact-dependent cell growth and proliferation [45–48]. YAP and TAZ, key downstream effectors of the Hippo pathway, are activated following damage to various organs, playing roles in tissue regeneration and wound healing [45]. The Hippo pathway has also been implicated in metastasis, with evidence showing that components such as LAT1/2, MST1/2, YAP, and TAZ influence tumor spread [45, 49]. In most contexts, Hippo signaling acts as an inhibitor of metastasis [5]. Another important role of this pathway is in treatment resistance, with studies suggesting that resistance to systemic therapies used in BC can be reversed by targeting Hippo pathway elements [45, 47]. The relationship between Hippo signaling and PIK3CAm may lie in the fact that YAP and TAZ drive epithelial cell proliferation in response to both mechanical stimuli and growth factor signaling through the PI3K-PDK1-Akt pathway [50]. Our observation of longer time to distant metastasis in patients with PIK3CAm may be explained by the lower frequency of Hippo pathway alterations in this group, given the pathway’s role in metastasis. A possible hypothesis for these lower rates of Hippo alteration is mutual exclusivity, the principle that alterations affecting the same biological pathway tend not to co-occur within the same patient [51]. Finally, these lower rates of Hippo alterations may also relate to resistance to HER2-targeted therapies in PIK3CAm, as inhibition of Hippo signaling has been shown to mitigate treatment resistance. From a clinical practice perspective, despite the evidence linking PIK3CAm with poor outcomes in HER2 + BC patients, and despite clear evidence that PIK3CAm drives resistance to standard HER2-targeted therapies by maintaining PI3K pathway activation, PIK3CA mutation testing isn’t currently routinely integrated into clinical decision-making for HER2 + patients, and no PI3K pathway inhibitors are currently FDA-approved specifically for HER2 + BC [15, 30, 41]. Given that PIK3CAm Stage IV HER2 + patients receiving first-line therapy had a lower PFS, our study reinforces that routine PIK3CA mutation testing at diagnosis for HER2 + patients could be explored as a potential tool to risk-stratify patients and guide future personalized treatment strategies. While currently approved PI3K inhibitors are primarily for HR+/HER2- BC, our real-world findings provide further rationale for evaluating such agents in clinical trials for HER2 + PIK3CAm patients who exhibit resistance or rapid progression on standard HER2-directed therapies. There are several ongoing clinical trials currently addressing this question (Supplemental Table 19). Our study has several limitations. The use of retrospective data can introduce issues of data quality, completeness, and confounding [52]. The GENIE BPC dataset has documented differences compared to other population-based registries [53, 54]. Studies have shown that patients in GENIE BPC colorectal cancer cohort tend to be younger, present with more advanced disease, have a greater prevalence of aggressive subtypes like triple-negative BC, and receive chemotherapy more frequently, often resulting in lower overall survival rates compared to SEER populations [54]. These inherent characteristics of the GENIE BPC cohort may limit the generalizability of our findings to the broader HER2 + BC population and necessitate careful interpretation when extrapolating results. The observation that some patients with PIK3CAm HER2 + BC did not receive standard anti-HER2 therapy likely reflects inter-institutional variability and provider-level differences, indicating that the data may contain heterogeneity and real-world nuances due to the multi-institutional nature of the dataset. Furthermore, the small number of patients in the PIK3CAm subgroup (n = 58) and subsequent subgroup analyses can reduce statistical power and lead to wide confidence intervals for certain estimates, potentially impacting the robustness of some conclusions. This may also explain the discrepancy in p-values observed between the log-rank and Cox regression analyses for DRFS. The limited sample sizes in several subgroup analyses (e.g., by line of therapy, stage, and ER status within the PIK3CAm group) resulted in insufficient patient numbers for robust multivariable modeling, precluding us from fully elucidating independent predictors in these specific contexts. Some patients began therapy prior to final diagnostic confirmation, which caused the lower median OS observed from diagnosis compared with OS from treatment initiation. On the other side, the use of a large, multi-institutional, real-world database like GENIE BPC provides valuable insights into clinical practice and complements findings from highly controlled clinical trials by reflecting actual patient management and outcomes in diverse clinical settings. The integration of comprehensive clinical and genomic data allows for a holistic evaluation of disease characteristics and treatment outcomes. The rigorous quality control and assurance protocols employed by GENIE BPC enhance the reliability of the data, and our use of advanced statistical methods like LASSO regression helped to manage collinearity and select key variables in a data-driven manner where sufficient power existed. In conclusion, our findings underscore the impact of PIK3CAm in HER2 + BC, especially in the metastatic setting. It could be explored as an important biomarker for identifying patients at a higher risk of disease progression and those who may be more resistant to standard HER2-targeted therapies. This necessitates the consideration of personalized treatment strategies, potentially including PI3K pathway inhibitors or alternative regimens, and emphasizes the need for enhanced monitoring for disease progression in this patient subgroup. Future prospective studies with larger, multicentric, more diverse cohorts are warranted to validate and expand these findings. Additionally, deeper bioinformatic mining of genomic data from Whole Exome/Genome Sequencing, starting with the raw Variant Call Format files, could help elucidate the underlying pathway-level mechanisms driving these clinical observations. Supplementary Information Supplementary Material 1 Acknowledgements The authors would like to acknowledge the American Association for Cancer Research and its financial and material support in the development of the AACR Project GENIE registry as well as members of the AACR Project GENIE consortium for their commitment to data sharing. Interpretations are the responsibility of study authors. Author contributions Study Planning: AJM and NS conceived the study concept, and the study design was by AJM, and NS.▪ Study Conduct : NS had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.▪ Study Reporting: NS, and AJM drafted the first version of the manuscript. All authors provided critical revision of the manuscript for important intellectual content. Study Planning: AJM and NS conceived the study concept, and the study design was by AJM, and NS. Study Conduct: NS had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study Reporting: NS, and AJM drafted the first version of the manuscript. All authors provided critical revision of the manuscript for important intellectual content. Funding This study did not receive any funding. Data availability Data from AACR Project GENIE are available upon request. Access can be obtained through the following platforms: (1) Synapse Platform (SageBionetworks): https://synapse.org/genie; (2) cBioPortal for Cancer Genomics (MSK): https://www.cbioportal.org/genie/. Declarations Ethical approval and consent to participate This study used deidentified data from the GENIE BPC database and was therefore exempt from institutional review board (IRB) review and the requirement for informed consent. Competing interests None of the authors has any competing interest. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Data from AACR Project GENIE are available upon request. Access can be obtained through the following platforms: (1) Synapse Platform (SageBionetworks): https://synapse.org/genie; (2) cBioPortal for Cancer Genomics (MSK): https://www.cbioportal.org/genie/.