Cell-of-Origin Patterns Dominate the Molecular Classification of 10,000 Tumors from 33 Types of Cancer.

We conducted comprehensive integrative molecular analyses of the complete set of tumors in The Cancer Genome Atlas (TCGA), consisting of approximately 10,000 specimens and representing 33 types of cancer. We performed molecular clustering using data on chromosome-arm-level aneuploidy, DNA hypermethylation, mRNA, and miRNA expression levels and reverse-phase protein arrays, of which all, except for aneuploidy, revealed clustering primarily organized by histology, tissue type, or anatomic origin. The influence of cell type was evident in DNA-methylation-based clustering, even after excluding sites with known preexisting tissue-type-specific methylation. Integrative clustering further emphasized the dominant role of cell-of-origin patterns. Molecular similarities among histologically or anatomically related cancer types provide a basis for focused pan-cancer analyses, such as pan-gastrointestinal, pan-gynecological, pan-kidney, and pan-squamous cancers, and those related by stemness features, which in turn may inform strategies for future therapeutic development.

Oncogenic Signaling Pathways in The Cancer Genome Atlas.

Genetic alterations in signaling pathways that control cell-cycle progression, apoptosis, and cell growth are common hallmarks of cancer, but the extent, mechanisms, and co-occurrence of alterations in these pathways differ between individual tumors and tumor types. Using mutations, copy-number changes, mRNA expression, gene fusions and DNA methylation in 9,125 tumors profiled by The Cancer Genome Atlas (TCGA), we analyzed the mechanisms and patterns of somatic alterations in ten canonical pathways: cell cycle, Hippo, Myc, Notch, Nrf2, PI-3-Kinase/Akt, RTK-RAS, TGFß signaling, p53 and ß-catenin/Wnt. We charted the detailed landscape of pathway alterations in 33 cancer types, stratified into 64 subtypes, and identified patterns of co-occurrence and mutual exclusivity. Eighty-nine percent of tumors had at least one driver alteration in these pathways, and 57% percent of tumors had at least one alteration potentially targetable by currently available drugs. Thirty percent of tumors had multiple targetable alterations, indicating opportunities for combination therapy.

Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer.

We report a comprehensive analysis of 412 muscle-invasive bladder cancers characterized by multiple TCGA analytical platforms. Fifty-eight genes were significantly mutated, and the overall mutational load was associated with APOBEC-signature mutagenesis. Clustering by mutation signature identified a high-mutation subset with 75% 5-year survival. mRNA expression clustering refined prior clustering analyses and identified a poor-survival "neuronal" subtype in which the majority of tumors lacked small cell or neuroendocrine histology. Clustering by mRNA, long non-coding RNA (lncRNA), and miRNA expression converged to identify subsets with differential epithelial-mesenchymal transition status, carcinoma in situ scores, histologic features, and survival. Our analyses identified 5 expression subtypes that may stratify response to different treatments.

Oncogenic IDH mutations increase heterochromatin-related replication stress without impacting homologous recombination.

Oncogenic mutations in isocitrate dehydrogenases 1 and 2 (IDH1/2) produce 2-hydroxyglutarate (2HG), which inhibits dioxygenases that modulate chromatin dynamics. The effects of 2HG have been reported to sensitize IDH tumors to poly-(ADP-ribose) polymerase (PARP) inhibitors. However, unlike PARP-inhibitor-sensitive BRCA1/2 tumors, which exhibit impaired homologous recombination, IDH-mutant tumors have a silent mutational profile and lack signatures associated with impaired homologous recombination. Instead, 2HG-producing IDH mutations lead to a heterochromatin-dependent slowing of DNA replication accompanied by increased replication stress and DNA double-strand breaks. This replicative stress manifests as replication fork slowing, but the breaks are repaired without a significant increase in mutation burden. Faithful resolution of replicative stress in IDH-mutant cells is dependent on poly-(ADP-ribosylation). Treatment with PARP inhibitors increases DNA replication but results in incomplete DNA repair. These findings demonstrate a role for PARP in the replication of heterochromatin and further validate PARP as a therapeutic target in IDH-mutant tumors.

Genomic profiling identifies somatic mutations predicting thromboembolic risk in patients with solid tumors.

Venous thromboembolism (VTE) associated with cancer (CAT) is a well-described complication of cancer and a leading cause of death in patients with cancer. The purpose of this study was to assess potential associations of molecular signatures with CAT, including tumor-specific mutations and the presence of clonal hematopoiesis. We analyzed deep-coverage targeted DNA-sequencing data of >14 000 solid tumor samples using the Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets platform to identify somatic alterations associated with VTE. End point was defined as the first instance of cancer-associated pulmonary embolism and/or proximal/distal lower extremity deep vein thrombosis. Cause-specific Cox proportional hazards regression was used, adjusting for pertinent clinical covariates. Of 11 695 evaluable individuals, 72% had metastatic disease at time of analysis. Tumor-specific mutations in KRAS (hazard ratio [HR], 1.34; 95% confidence interval (CI), 1.09-1.64; adjusted P = .08), STK11 (HR, 2.12; 95% CI, 1.55-2.89; adjusted P < .001), KEAP1 (HR, 1.84; 95% CI, 1.21-2.79; adjusted P = .07), CTNNB1 (HR, 1.73; 95% CI, 1.15-2.60; adjusted P = .09), CDKN2B (HR, 1.45; 95% CI, 1.13-1.85; adjusted P = .07), and MET (HR, 1.83; 95% CI, 1.15-2.92; adjusted P = .09) were associated with a significantly increased risk of CAT independent of tumor type. Mutations in SETD2 were associated with a decreased risk of CAT (HR, 0.35; 95% CI, 0.16-0.79; adjusted P = .09). The presence of clonal hematopoiesis was not associated with an increased VTE rate. This is the first large-scale analysis to elucidate tumor-specific genomic events associated with CAT. Somatic tumor mutations of STK11, KRAS, CTNNB1, KEAP1, CDKN2B, and MET were associated with an increased risk of VTE in patients with solid tumors. Further analysis is needed to validate these findings and identify additional molecular signatures unique to individual tumor types.

Computational methods and translational applications for targeted next-generation sequencing platforms.

During the past decade, next-generation sequencing (NGS) technologies have become widely adopted in cancer research and clinical care. Common applications within the clinical setting include patient stratification into relevant molecular subtypes, identification of biomarkers of response and resistance to targeted and systemic therapies, assessment of heritable cancer risk based on known pathogenic variants, and longitudinal monitoring of treatment response. The need for efficient downstream processing and reliable interpretation of sequencing data has led to the development of novel algorithms and computational pipelines, as well as structured knowledge bases that link genomic alterations to currently available drugs and ongoing clinical trials. Cancer centers around the world use different types of targeted solid-tissue and blood based NGS assays to analyze the genomic and transcriptomic profile of patients as part of their routine clinical care. Recently, cross-institutional collaborations have led to the creation of large pooled datasets that can offer valuable insights into the genomics of rare cancers.

Adoption of Organ Preservation and Surgeon Variability for Patients with Rectal Cancer Does Not Correlate with Worse Survival.

BACKGROUND AND OBJECTIVES: Watch-and-wait is variably adopted by surgeons and the impact of this on outcomes is unknown. We compared the disease-free survival and organ preservation rates of locally advanced rectal cancer patients treated by expert colorectal surgeons at a comprehensive cancer center. METHODS: This study included retrospective data on patients diagnosed with stage II/III rectal adenocarcinoma from January 2013 to June 2017 who initiated neoadjuvant therapy (either with chemoradiation, chemotherapy, or a combination of both) and were treated by an expert colorectal surgeon. RESULTS: Overall, 444 locally advanced rectal cancer patients managed by five surgeons were included. Tumor distance from the anal verge, type of neoadjuvant therapy, and organ preservation rates varied by treating surgeon. There was no difference in disease-free survival after stratifying by the treating surgeon (p = 0.2). On multivariable analysis, neither the type of neoadjuvant therapy nor the treating surgeon was associated with disease-free survival. CONCLUSIONS: While neoadjuvant therapy type and organ preservation rates varied among surgeons, there were no meaningful differences in disease-free survival. These data suggest that among expert colorectal surgeons, differing thresholds for selecting patients for watch-and-wait do not affect survival.

Genomic Predictors of Recurrence Patterns After Complete Resection of Colorectal Liver Metastases and Adjuvant Hepatic Artery Infusion Chemotherapy.

BACKGROUND: Despite curative hepatectomy, most colorectal liver metastasis (CRLM) patients relapse locally within 2 years. Genomic predictors for hepatic recurrence are poorly understood. This study was designed to identify genomic signatures for recurrence in resected CRLM patients treated with adjuvant hepatic artery infusion (HAI) and/or systemic (SYS) chemotherapy. METHODS: Patients undergoing curative hepatectomy and adjuvant HAI+SYS or SYS between January 2000 and October 2017 with next-generation sequencing data were catalogued. Gene and signaling-level alterations were checked for association with time to any (AR), liver (LR), and extrahepatic recurrence (ER) by using Kaplan-Meier analysis. RESULTS: Of 172 receiving HAI+SYS, 100 patients recurred, with 69 LR and 83 ER. Five- and ten-year LR-free rates were 57% (95% confidence interval [CI] 48-65%) and 51% (95% CI 41-60%), respectively. Five- and 10-year ER-free, rates were 51% (95% CI 43-58%) and 45% (95% CI 36-54%), respectively. More ER was observed with tumors harboring altered KRAS (38% [95% CI 25-50%] vs. 63% [95% CI 53-71%], p-adj = 0.003) and RAS/RAF (36% [95% CI 25-48%] vs. 66% [95% CI 56-74%], p-adj < 0.001) than wild-type. Co-altered RAS/RAF-TP53 was associated with worse AR (26% [95% CI 14-40%] vs. 48% [95% CI 39-57%], p-unadj < 0.001), ER (30% [95% CI 17-45%] vs. 62% [95% CI 53-70%], p-unadj < 0.001), and LR rate (40% [95% CI 24-57%] vs. 70% [95% CI 60-77%], p-unadj = 0.002). On multivariable analysis, controlling for clinical risk score, ablation, margin status, and primary T-stage, co-altered RAS/RAF-TP53 was associated with increased risk for AR (HR = 2.14, 95% CI 1.38-3.31, p-unadj < 0.001), LR (HR = 1.79, 95% CI 1.06-3.02, p-unadj = 0.029), and ER (HR = 2.81, 95% CI 1.78-4.44, p-unadj < 0.001). CONCLUSIONS: Altered KRAS, RAS/RAF, and RAS/RAF-TP53 associated with earlier local and distant recurrence in resected CRLM patients receiving adjuvant HAI+SYS. Co-altered RAS/RAF-TP53 was a novel predictor of LR warranting investigation of whether genomic cooperativity is associated with this relapsing phenotype. Systemic therapies tailored to high-risk tumor biology are needed to reduce distant relapse after hepatectomy.

Association of genomic profiles and survival in early onset and screening-age colorectal cancer patients with liver metastases resected over 15 years.

BACKGROUND: This study explores whether genomic profiles of colorectal liver metastasis (CRLM) patients with early onset (EO, < 50 years old) and screening age (SA) primary diagnosis are associated with overall survival (OS). METHODS: All patients undergoing hepatectomy between 2002 and 2017 were identified and tumor specimens with next-generation sequencing data were cataloged. Gene and signaling-level alterations were checked for association with OS from primary diagnosis accommodating for left-truncated survival. RESULTS: Of 1822 patients, 333 were sequenced-127 (38%) EO-CRLM and 206 (62%) SA-CRLM patients. More aggressive features presented in EO-CRLM patients-synchronous metastatic presentation (83% vs. 75%, p < 0.001) and primary node-positive disease (71% vs. 61%, p < 0.001). The median OS from primary diagnosis was 11.8 years (95% confidence interval = 7.94-NA). Five-year OS did not differ by age (p = 0.702). On multivariable analysis, altered APC (EO-CRLM: [hazard ratio [HR] = 0.37, p = 0.018] vs. SA-CRLM:[HR = 0.61, p = 0.260]), BRAF (EO-CRLM:[HR = 4.38, p = 0.007] vs. SA-CRLM:[HR = 4.78, p = 0.032]), and RAS-TP53 (EO-CRLM:[HR = 2.82, p = 0.011] vs. SA-CRLM:[HR = 2.35, p = 0.003]) associated with OS. CONCLUSIONS: Despite bearing more aggressive features, EO-CRLM patients had similar genomic profiles and survival as SA-CRLM patients. Better performance status in younger patients leading to increased treatment tolerance may partly explain this. As screening and treatment strategies from older patients are applied to younger patients, genomic predictors of biology identified historically in older cohorts could apply to EO patients.

Intraoperative opioids are associated with decreased recurrence rates in colon adenocarcinoma: a retrospective observational cohort study.

BACKGROUND: Opioid-induced immunomodulation may be important in colon adenocarcinoma, where tumour DNA mismatch repair (MMR) can determine the level of immune activation with consequences for therapeutic response and prognosis. We evaluated the relationship between intraoperative opioid exposure, MMR subtype, and oncological outcomes after surgery for colon adenocarcinoma. METHODS: Intraoperative opioid use (standardised by calculating morphine milligram equivalents) during stage I-III colon adenocarcinoma resection was reviewed retrospectively. Tumours were classified as DNA mismatch repair deficient (dMMR) or proficient (pMMR) by immunohistochemistry. The primary outcome was local tumour recurrence, distant tumour recurrence, or both (multivariable analysis). The exposures of interest were intraoperative analgesia and tumour subtype. Opioid-related gene expression was analysed using The Cancer Genome Atlas Colon Adenocarcinoma transcriptomic data. RESULTS: Clinical and pathological data were analysed from 1157 subjects (median age, 60 [51-70] yr; 49% female) who underwent curative resection for stage I-III colon adenocarcinoma. Higher intraoperative opioid doses were associated with reduced risk of tumour recurrence (hazard ratio=0.92 per 10 morphine milligram equivalents; 95% confidence interval [95% CI], 0.87-0.98; P=0.007), but not with overall survival. In tumours deficient in DNA MMR, tumour recurrence was less likely (HR=0.38; 95% CI, 0.21-0.68; P=0.001), with higher opioid dose associated with eightfold lower recurrence rates. Gene expression related to opioid signalling was different between dMMR and pMMR tumours. CONCLUSIONS: Higher intraoperative opioid dose was associated with a lower risk of tumour recurrence after surgery for stage I-III colon adenocarcinoma, but particularly so in tumours in which DNA MMR was deficient.

A Randomized Phase II Trial of Adjuvant Hepatic Arterial Infusion and Systemic Therapy With or Without Panitumumab After Hepatic Resection of KRAS Wild-type Colorectal Cancer.

OBJECTIVE/BACKGROUND: The purpose was to determine whether adding Pmab versus no Pmab to an adjuvant regimen of hepatic arterial infusion (HAI) of floxuridine (FUDR) plus systemic (SYS) leucovorin, fluorouracil, and irinotecan (FOLFIRI) improves 15-month recurrence-free survival for patients with RAS wild-type colorectal cancer. Secondary endpoints included overall survival, toxicity, and influence of predictive biomarkers. METHODS: This phase II trial randomized patients with KRAS wild-type resected colorectal liver metastases to adjuvant HAI FUDR + SYS FOLFIRI +/- Pmab (NCT01312857). Patients were stratified by clinical risk score and previous chemotherapy. Based on an exact binomial design, if one arm had ≥24 patients alive and disease-free at 15 months that regimen was considered promising for further investigation. RESULTS: Seventy-five patients were randomized. Patient characteristics and toxicity were not different in the 2 arms, except for rash in +Pmab arm. Grade 3/4 elevation in bilirubin or alkaline phosphatase did not differ in the 2 arms. Twenty-five (69%; 95% CI, 53-82) patients in the Pmab arm versus 18 (47%; 95% CI, 32-63) patients in the arm without Pmab were alive and recurrence-free at 15 months. Only the Pmab arm met the decision rule, while the other arm did not. After median follow-up of 56.6 months, 3-year recurrence-free survival was 57% (95% CI, 43-76) and 42% (95% CI, 29-61), and 3-year overall survival was 97% (95% CI, 90-99) and 91% (95% CI, 83-99), +/- Pmab, respectively. CONCLUSIONS: The addition of Pmab to HAI FUDR + SYS FOLFIRI showed promising activity without increased biliary toxicity and should be further investigated in a larger trial.

Intraoperative opioid exposure, tumour genomic alterations, and survival differences in people with lung adenocarcinoma.

BACKGROUND: Opioids have been linked to worse oncologic outcomes in surgical patients. Studies in certain cancer types have identified associations between survival and intra-tumoural opioid receptor gene alterations, but no study has investigated whether the tumour genome interacts with opioid exposure to affect survival. We sought to determine whether intraoperative opioid exposure is associated with recurrence-specific survival and overall survival in early-stage lung adenocarcinoma, and whether selected tumour genomics are associated with this relationship. Associations between ketamine and dexmedetomidine and outcomes were also studied. METHODS: Surgical patients (N=740) with pathological stage I-III lung adenocarcinoma and next-generation sequencing data were retrospectively reviewed from a prospectively maintained database. RESULTS: On multivariable analysis, ketamine administration was protective for recurrence-specific survival (hazard ratio = 0.44, 95% confidence interval 0.24-0.80; P=0.007), compared with no adjunct. Higher intraoperative oral morphine milligram equivalents were significantly associated with worse overall survival (hazard ratio=1.09/10 morphine milligram equivalents, 95% confidence interval 1.02-1.17; P=0.010). Significant interaction effects were found between morphine milligram equivalents and fraction genome altered and morphine milligram equivalents and CDKN2A, such that higher fraction genome altered or CDKN2A alterations were associated with worse overall survival at higher morphine milligram equivalents (P=0.044 and P=0.052, respectively). In contrast, alterations in the Wnt (P=0.029) and Hippo (P=0.040) oncogenic pathways were associated with improved recurrence-specific survival at higher morphine milligram equivalents, compared with unaltered pathways. CONCLUSIONS: Intraoperative opioid exposure is associated with worse overall survival, whereas ketamine exposure is associated with improved recurrence-specific survival in patients with early-stage lung adenocarcinoma. This is the first study to investigate tumour-specific genomic interactions with intraoperative opioid administration to modify survival associations.

Lung-only melanoma: UV mutational signature supports origin from occult cutaneous primaries and argues against the concept of primary pulmonary melanoma.

Primary pulmonary melanoma (PPM) is an entity recognized by the thoracic WHO classification. However, given the absence of native melanocytes in the lung and the known phenomenon of regression of cutaneous melanomas, the existence of PPM has remained controversial. Herein we investigate clinicopathologic and genomic features of lung-only melanomas with the goal to clarify their site of origin. We identified 10 melanomas involving exclusively lung with no current or previous cutaneous, uveal, or mucosal primaries. Four patients had solitary lesions with mean size of 5.1 cm (range 3.0-10.1 cm), meeting the criteria of PPM. Four patients had 2-3 lesions and 2 patients had >10 lesions. All cases underwent targeted next-generation sequencing interrogating up to 468 cancer genes, which revealed mean tumor mutation burden of 42.6 per megabase (range 1.8 to 126) and frequent mutations involving BRAF, NRAS, NF1, KIT, and KRAS - a genomic profile typical of UV-associated cutaneous melanoma. Mutational signature was assessable for eight cases harboring >20 mutations. This revealed that all evaluable cases harbored a dominant UV signature. In addition, one nonevaluable case harbored a GG > AA TERT promoter variant that is highly specific for UV-mutagenesis. As control groups, using the same methodology, a dominant UV signature was identified in 97% (470/486) of cutaneous melanomas, whereas no lung adenocarcinoma (n = 291) exhibited this signature. Notably, the clinical and pathologic features of solitary melanomas, especially those with large size and epithelioid morphology, closely mimicked primary lung carcinomas, highlighting a major potential for misdiagnosis. In conclusion, presence of a UV signature provides direct evidence that nearly all lung-only melanomas in this series, including solitary lesions meeting the strict criteria of PPM, represent metastases from occult cutaneous melanomas. This suggests that lung-only melanomas should be considered as likely metastatic even in the absence of a known primary melanoma elsewhere.

Transposon mutagenesis identifies chromatin modifiers cooperating with Ras in thyroid tumorigenesis and detects ATXN7 as a cancer gene.

Oncogenic RAS mutations are present in 15-30% of thyroid carcinomas. Endogenous expression of mutant Ras is insufficient to initiate thyroid tumorigenesis in murine models, indicating that additional genetic alterations are required. We used Sleeping Beauty (SB) transposon mutagenesis to identify events that cooperate with HrasG12V in thyroid tumor development. Random genomic integration of SB transposons primarily generated loss-of-function events that significantly increased thyroid tumor penetrance in Tpo-Cre/homozygous FR-HrasG12V mice. The thyroid tumors closely phenocopied the histological features of human RAS-driven, poorly differentiated thyroid cancers. Characterization of transposon insertion sites in the SB-induced tumors identified 45 recurrently mutated candidate cancer genes. These mutation profiles were remarkably concordant with mutated cancer genes identified in a large series of human poorly differentiated and anaplastic thyroid cancers screened by next-generation sequencing using the MSK-IMPACT panel of cancer genes, which we modified to include all SB candidates. The disrupted genes primarily clustered in chromatin remodeling functional nodes and in the PI3K pathway. ATXN7, a component of a multiprotein complex with histone acetylase activity, scored as a significant SB hit. It was recurrently mutated in advanced human cancers and significantly co-occurred with RAS or NF1 mutations. Expression of ATXN7 mutants cooperated with oncogenic RAS to induce thyroid cell proliferation, pointing to ATXN7 as a previously unrecognized cancer gene.

Phase II study of trastuzumab with modified docetaxel, cisplatin, and 5 fluorouracil in metastatic HER2-positive gastric cancer.

BACKGROUND: Trastuzumab with cisplatin and fluoropyrimidine is the standard treatment in metastatic HER2-positive gastric or gastroesophageal (GE) junction adenocarcinoma; however, there is limited data on the efficacy of trastuzumab in combination with a three-drug regimen in this setting. We examined the efficacy and safety of modified docetaxel, cisplatin and 5 fluorouracil (mDCF) plus trastuzumab in a single-arm multicenter phase II trial. METHODS: Previously untreated patients with HER2-positive metastatic gastric or GE junction adenocarcinoma were treated with mDCF and trastuzumab every 2 weeks. The primary endpoint was 6-month progression-free survival (PFS); secondary endpoints included objective response rate, overall survival (OS), and toxicity. RESULTS: We enrolled 26 patients with metastatic HER2-positive gastric or GE junction adenocarcinoma between February 2011 and June 2015. The median age of patients was 62 years; 96% had a Karnofsky performance status equal to or greater than 80%. With a median follow-up of 25.4 months, the 6-month PFS was 73% (95% CI 51-86%). The objective response rate was 65%, the median PFS was 13 months (95% CI 6.4-20.7) and the median OS was 24.9 months (95% CI 14.4-42.5). Grade 3/4 toxicities included neutropenia (42%), fatigue (23%), and hypophosphatemia (15%). There were no episodes of febrile neutropenia. CONCLUSION: The combination of mDCF and trastuzumab is effective and safe in patients with metastatic HER2-positive gastric or GE junction adenocarcinoma and can be considered as an option for selected patients. This trial is registered at ClinicalTrials.gov, number NCT00515411.

A Multi-Method Approach for Proteomic Network Inference in 11 Human Cancers.

Protein expression and post-translational modification levels are tightly regulated in neoplastic cells to maintain cellular processes known as 'cancer hallmarks'. The first Pan-Cancer initiative of The Cancer Genome Atlas (TCGA) Research Network has aggregated protein expression profiles for 3,467 patient samples from 11 tumor types using the antibody based reverse phase protein array (RPPA) technology. The resultant proteomic data can be utilized to computationally infer protein-protein interaction (PPI) networks and to study the commonalities and differences across tumor types. In this study, we compare the performance of 13 established network inference methods in their capacity to retrieve the curated Pathway Commons interactions from RPPA data. We observe that no single method has the best performance in all tumor types, but a group of six methods, including diverse techniques such as correlation, mutual information, and regression, consistently rank highly among the tested methods. We utilize the high performing methods to obtain a consensus network; and identify four robust and densely connected modules that reveal biological processes as well as suggest antibody-related technical biases. Mapping the consensus network interactions to Reactome gene lists confirms the pan-cancer importance of signal transduction pathways, innate and adaptive immune signaling, cell cycle, metabolism, and DNA repair; and also suggests several biological processes that may be specific to a subset of tumor types. Our results illustrate the utility of the RPPA platform as a tool to study proteomic networks in cancer.

Weakly Supervised Deep Learning Predicts Immunotherapy Response in Solid Tumors Based on PD-L1 Expression.

Programmed death-ligand 1 (PD-L1) IHC is the most commonly used biomarker for immunotherapy response. However, quantification of PD-L1 status in pathology slides is challenging. Neither manual quantification nor a computer-based mimicking of manual readouts is perfectly reproducible, and the predictive performance of both approaches regarding immunotherapy response is limited. In this study, we developed a deep learning (DL) method to predict PD-L1 status directly from raw IHC image data, without explicit intermediary steps such as cell detection or pigment quantification. We trained the weakly supervised model on PD-L1-stained slides from the non-small cell lung cancer (NSCLC)-Memorial Sloan Kettering (MSK) cohort (N = 233) and validated it on the pan-cancer-Vall d'Hebron Institute of Oncology (VHIO) cohort (N = 108). We also investigated the performance of the model to predict response to immune checkpoint inhibitors (ICI) in terms of progression-free survival. In the pan-cancer-VHIO cohort, the performance was compared with tumor proportion score (TPS) and combined positive score (CPS). The DL model showed good performance in predicting PD-L1 expression (TPS ≥ 1%) in both NSCLC-MSK and pan-cancer-VHIO cohort (AUC 0.88 ± 0.06 and 0.80 ± 0.03, respectively). The predicted PD-L1 status showed an improved association with response to ICIs [HR: 1.5 (95% confidence interval: 1-2.3), P = 0.049] compared with TPS [HR: 1.4 (0.96-2.2), P = 0.082] and CPS [HR: 1.2 (0.79-1.9), P = 0.386]. Notably, our explainability analysis showed that the model does not just look at the amount of brown pigment in the IHC slides, but also considers morphologic factors such as lymphocyte conglomerates. Overall, end-to-end weakly supervised DL shows potential for improving patient stratification for cancer immunotherapy by analyzing PD-L1 IHC, holistically integrating morphology and PD-L1 staining intensity. SIGNIFICANCE: The weakly supervised DL model to predict PD-L1 status from raw IHC data, integrating tumor staining intensity and morphology, enables enhanced patient stratification in cancer immunotherapy compared with traditional pathologist assessment.

Oncologic Outcomes of Salvage Abdominoperineal Resection for Anal Squamous Cell Carcinoma Initially Managed with Chemoradiation.

Background: Abdominoperineal resection (APR) has been advocated for persistent or recurrent disease after failure of chemoradiation (CRT) for anal squamous cell cancer (SCC). Treatment with salvage APR can potentially achieve a cure. This study aimed to analyze oncological outcomes for salvage APR in a recent time period at a comprehensive cancer center. Methods: A retrospective review of all patients who underwent APR for biopsy-proven persistent or recurrent anal SCC between 1 January 2007 and 31 December 2020 was performed. Patients with stage IV disease at the time of initial diagnosis and patients with missing data were excluded. Univariate analysis was used with a chi-square test for categorical variables, and non-parametric tests were used for continuous variables. Kaplan-Meier survival analysis was performed to evaluate disease-specific (DSS), post-APR local recurrence-free (RFS), and disease-free survival (DFS). Results: A total of 96 patients were included in the analysis: 39 (41%) with persistent disease and 57 (59%) with recurrent SCC after chemoradiation had been completed. The median follow-up was 22 months (IQR 11-47). Forty-nine patients (51%) underwent extended APR and/or pelvic exenteration. Eight (8%) patients developed local recurrence, 30 (31%) developed local and distant recurrences, and 16 (17%) developed distant recurrences alone. The 3-year DSS, post-APR local recurrence-free survival, and disease-free survival were 53.8% (95% CI 43.5-66.5%), 54.5% (95% CI 44.4-66.8%), and 26.8% (95% CI 18.6-38.7%), respectively. In multivariate logistic regression analysis, positive microscopic margin (OR 10.0, 95% CI 2.16-46.12, p = 0.003), positive nodes in the surgical specimen (OR 9.19, 95% CI 1.99-42.52, p = 0.005), and lymphovascular invasion (OR 2.61 95% CI 1.05-6.51, p = 0.04) were associated with recurrence of disease. Gender, indication for APR (recurrent vs. persistent disease), HIV status, extent of surgery, or type of reconstruction did not influence survival outcomes. Twenty patients had targeted tumor-sequencing data available. Nine patients had PIK3CA mutations, seven of whom experienced a recurrence. Conclusions: Salvage APR for anal SCC after failed CRT was associated with poor disease-specific survival and low recurrence-free survival. Anal SCC patients undergoing salvage APR should be counseled that microscopic positive margins, positive lymph nodes, or the presence of lymphovascular invasion in the APR specimen are prognosticators for disease relapse. Our results accentuate the necessity for additional treatment strategies for the ongoing treatment challenge of persistent or recurrent anal SCC after failed CRT.