Generalization optimizing machine learning to improve CT scan radiomics and assess immune checkpoint inhibitors' response in non-small cell lung cancer: a multicenter cohort study.

Background: Recent developments in artificial intelligence suggest that radiomics may represent a promising non-invasive biomarker to predict response to immune checkpoint inhibitors (ICIs). Nevertheless, validation of radiomics algorithms in independent cohorts remains a challenge due to variations in image acquisition and reconstruction. Using radiomics, we investigated the importance of scan normalization as part of a broader machine learning framework to enable model external generalizability to predict ICI response in non-small cell lung cancer (NSCLC) patients across different centers. Methods: Radiomics features were extracted and compared from 642 advanced NSCLC patients on pre-ICI scans using established open-source PyRadiomics and a proprietary DeepRadiomics deep learning technology. The population was separated into two groups: a discovery cohort of 512 NSCLC patients from three academic centers and a validation cohort that included 130 NSCLC patients from a fourth center. We harmonized images to account for variations in reconstruction kernel, slice thicknesses, and device manufacturers. Multivariable models, evaluated using cross-validation, were used to estimate the predictive value of clinical variables, PD-L1 expression, and PyRadiomics or DeepRadiomics for progression-free survival at 6 months (PFS-6). Results: The best prognostic factor for PFS-6, excluding radiomics features, was obtained with the combination of Clinical + PD-L1 expression (AUC = 0.66 in the discovery and 0.62 in the validation cohort). Without image harmonization, combining Clinical + PyRadiomics or DeepRadiomics delivered an AUC = 0.69 and 0.69, respectively, in the discovery cohort, but dropped to 0.57 and 0.52, in the validation cohort. This lack of generalizability was consistent with observations in principal component analysis clustered by CT scan parameters. Subsequently, image harmonization eliminated these clusters. The combination of Clinical + DeepRadiomics reached an AUC = 0.67 and 0.63 in the discovery and validation cohort, respectively. Conversely, the combination of Clinical + PyRadiomics failed generalizability validations, with AUC = 0.66 and 0.59. Conclusion: We demonstrated that a risk prediction model combining Clinical + DeepRadiomics was generalizable following CT scan harmonization and machine learning generalization methods. These results had similar performances to routine oncology practice using Clinical + PD-L1. This study supports the strong potential of radiomics as a future non-invasive strategy to predict ICI response in advanced NSCLC.

Interplay of Immunosuppression and Immunotherapy Among Patients With Cancer and COVID-19.

Importance: Cytokine storm due to COVID-19 can cause high morbidity and mortality and may be more common in patients with cancer treated with immunotherapy (IO) due to immune system activation. Objective: To determine the association of baseline immunosuppression and/or IO-based therapies with COVID-19 severity and cytokine storm in patients with cancer. Design, Setting, and Participants: This registry-based retrospective cohort study included 12 046 patients reported to the COVID-19 and Cancer Consortium (CCC19) registry from March 2020 to May 2022. The CCC19 registry is a centralized international multi-institutional registry of patients with COVID-19 with a current or past diagnosis of cancer. Records analyzed included patients with active or previous cancer who had a laboratory-confirmed infection with SARS-CoV-2 by polymerase chain reaction and/or serologic findings. Exposures: Immunosuppression due to therapy; systemic anticancer therapy (IO or non-IO). Main Outcomes and Measures: The primary outcome was a 5-level ordinal scale of COVID-19 severity: no complications; hospitalized without requiring oxygen; hospitalized and required oxygen; intensive care unit admission and/or mechanical ventilation; death. The secondary outcome was the occurrence of cytokine storm. Results: The median age of the entire cohort was 65 years (interquartile range [IQR], 54-74) years and 6359 patients were female (52.8%) and 6598 (54.8%) were non-Hispanic White. A total of 599 (5.0%) patients received IO, whereas 4327 (35.9%) received non-IO systemic anticancer therapies, and 7120 (59.1%) did not receive any antineoplastic regimen within 3 months prior to COVID-19 diagnosis. Although no difference in COVID-19 severity and cytokine storm was found in the IO group compared with the untreated group in the total cohort (adjusted odds ratio [aOR], 0.80; 95% CI, 0.56-1.13, and aOR, 0.89; 95% CI, 0.41-1.93, respectively), patients with baseline immunosuppression treated with IO (vs untreated) had worse COVID-19 severity and cytokine storm (aOR, 3.33; 95% CI, 1.38-8.01, and aOR, 4.41; 95% CI, 1.71-11.38, respectively). Patients with immunosuppression receiving non-IO therapies (vs untreated) also had worse COVID-19 severity (aOR, 1.79; 95% CI, 1.36-2.35) and cytokine storm (aOR, 2.32; 95% CI, 1.42-3.79). Conclusions and Relevance: This cohort study found that in patients with cancer and COVID-19, administration of systemic anticancer therapies, especially IO, in the context of baseline immunosuppression was associated with severe clinical outcomes and the development of cytokine storm. Trial Registration: ClinicalTrials.gov Identifier: NCT04354701.

Helicobacter pylori serology is associated with worse overall survival in patients with melanoma treated with immune checkpoint inhibitors.

The microbiome is now regarded as one of the hallmarks of cancer and several strategies to modify the gut microbiota to improve immune checkpoint inhibitor (ICI) activity are being evaluated in clinical trials. Preliminary data regarding the upper gastro-intestinal microbiota indicated that Helicobacter pylori seropositivity was associated with a negative prognosis in patients amenable to ICI. In 97 patients with advanced melanoma treated with ICI, we assessed the impact of H. pylori on outcomes and microbiome composition. We performed H. pylori serology and profiled the fecal microbiome with metagenomics sequencing. Among the 97 patients, 22% were H. pylori positive (Pos). H. pylori Pos patients had a significantly shorter overall survival (p = .02) compared to H. pylori negative (Neg) patients. In addition, objective response rate and progression-free survival were decreased in H. pylori Pos patients. Metagenomics sequencing did not reveal any difference in diversity indexes between the H. pylori groups. At the taxa level, Eubacterium ventriosum, Mediterraneibacter (Ruminococcus) torques, and Dorea formicigenerans were increased in the H. pylori Pos group, while Alistipes finegoldii, Hungatella hathewayi and Blautia producta were over-represented in the H. pylori Neg group. In a second independent cohort of patients with NSCLC, diversity indexes were similar in both groups and Bacteroides xylanisolvens was increased in H. pylori Neg patients. Our results demonstrated that the negative impact of H. pylori on outcomes seem to be independent from the fecal microbiome composition. These findings warrant further validation and development of therapeutic strategies to eradicate H. pylori in immuno-oncology arena.

Adaptive metabolic rewiring to chronic SFK inhibition.

Src family kinases (SFK) are key regulators of cellular proliferation, differentiation, survival, motility and angiogenesis. As such, SFK inhibitors are being tested in clinical trials to prevent metastasis as an alternative to current treatment regimens for a variety of cancers including breast cancer. To contribute to the development of molecular tools improving SFK-targeted therapies, we used the SFK inhibitor dasatinib and a well characterized triple negative breast cancer cell line (BT20). Comparison of the response of BT20 cells with acquired resistance to dasatinib and its' parental counterpart suggest that chronic exposure to SFK inhibition results in increased dependency on TGFß signaling for proliferation, both in the absence or the presence of dasatinib. In addition, we found that acquired (but not de novo) resistance to dasatinib was reduced by non-cytotoxic concentrations compounds hindering on PI3K, mTORC1 signaling, endoplasmic reticulum stress or autophagy.

Simultaneous inhibition of ATR and PARP sensitizes colon cancer cell lines to irinotecan.

Enhanced DNA damage repair is one mechanism involved in colon cancer drug resistance. Thus, targeting molecular components of repair pathways with specific small molecule inhibitors may improve the efficacy of chemotherapy. ABT-888 and VE-821, inhibitors of poly-ADP-ribose-polymerase (PARP) and the serine/threonine-kinase Ataxia telangiectasia related (ATR), respectively, were used to treat colon cancer cell lines in combination with the topoisomerase-I inhibitor irinotecan (SN38). Our findings show that each of these DNA repair inhibitors utilized alone at nontoxic single agent concentrations resulted in sensitization to SN38 producing a 1.4-3 fold reduction in the 50% inhibitory concentration (IC50) of SN38 in three colon cancer cell lines. When combined together, nontoxic concentrations of ABT-888 and VE-821 produced a 4.5-27 fold reduction in the IC50 of SN38 with the HCT-116 colon cancer cells demonstrating the highest sensitization as compared to LoVo and HT-29 colon cancer cells. Furthermore, the combination of all three agents was associated with maximal G2 -M arrest and enhanced DNA-damage (γH2AX) in all three colon cancer cell lines. The mechanism of this enhanced sensitization was associated with: (a) maximal suppression of SN38 induced PARP activity in the presence of both inhibitors and (b) ABT-888 producing partial abrogation of the VE-821 enhancement of SN38 induced DNA-PK phosphorylation, resulting in more unrepaired DNA damage; these alterations were only present in the HCT-116 cells which have reduced levels of ATM. This novel combination of DNA repair inhibitors may be useful to enhance the activity of DNA damaging chemotherapies such as irinotecan and help produce sensitization to this drug in colon cancer.