Correlation between biomarkers and treatment outcomes in diverse cancers: a systematic review and meta-analysis of phase I and II immunotherapy clinical trials

Convergence of evolving artificial intelligence and machine learning techniques in precision oncology

The confluence of new technologies with artificial intelligence (AI) and machine learning (ML) analytical techniques is rapidly advancing the field of precision oncology, promising to improve diagnostic approaches and therapeutic strategies for patients with cancer. By analyzing multi-dimensional, multiomic, spatial pathology, and radiomic data, these technologies enable a deeper understanding of the intricate molecular pathways, aiding in the identification of critical nodes within the tumor's biology to optimize treatment selection. The applications of AI/ML in precision oncology are extensive and include the generation of synthetic data, e.g., digital twins, in order to provide the necessary information to design or expedite the conduct of clinical trials. Currently, many operational and technical challenges exist related to data technology, engineering, and storage; algorithm development and structures; quality and quantity of the data and the analytical pipeline; data sharing and generalizability; and the incorporation of these technologies into the current clinical workflow and reimbursement models.

A multivariate Polya tree model for meta-analysis with event-time distributions

We develop a nonparametric Bayesian prior for a family of random probability measures by extending the Polya tree ($\mbox{PT}$) prior to a joint prior for a set of probability measures $G_1,\dots ,G_n$, suitable for meta-analysis with event-time outcomes. In the application to meta-analysis, $G_i$ is the event-time distribution specific to study $i$. The proposed model defines a regression on study-specific covariates by introducing increased correlation for any pair of studies with similar characteristics. The desired multivariate $\mbox{PT}$ model is constructed by introducing a hierarchical prior on the conditional splitting probabilities in the $\mbox{PT}$ construction for each of the $G_i$. The hierarchical prior replaces the independent beta priors for the splitting probability in the PT construction with a Gaussian process prior for corresponding (logit) splitting probabilities across all studies. The Gaussian process is indexed by study-specific covariates, introducing the desired dependence with increased correlation for similar studies. The main feature of the proposed construction is (conditionally) conjugate posterior updating with commonly reported inference summaries for event-time data. The construction is motivated by a meta-analysis over cancer immunotherapy studies.

OX40/OX40 ligand and its role in precision immune oncology

Immune checkpoint inhibitors have changed the treatment landscape for various malignancies; however, their benefit is limited to a subset of patients. The immune machinery includes both mediators of suppression/immune evasion, such as PD-1, PD-L1, CTLA-4, and LAG-3, all of which can be inhibited by specific antibodies, and immune-stimulatory molecules, such as T-cell co-stimulatory receptors that belong to the tumor necrosis factor receptor superfamily (TNFRSF), including OX40 receptor (CD134; TNFRSF4), 4-1BB (CD137; TNFRSF9), and glucocorticoid-induced TNFR-related (GITR) protein (CD357; TNFRSF18). In particular, OX40 and its binding ligand OX40L (CD134L; TNFSF4; CD252) are critical for immunoregulation. When OX40 on activated T cells binds OX40L on antigen-presenting cells, T-cell activation and immune stimulation are initiated via enhanced T-cell survival, proliferation and cytotoxicity, memory T-cell formation, and abrogation of regulatory T cell (Treg) immunosuppressive functions. OX40 agonists are in clinical trials both as monotherapy and in combination with other immunotherapy agents, in particular specific checkpoint inhibitors, for cancer treatment. To date, however, only a minority of patients respond. Transcriptomic profiling reveals that OX40 and OX40L expression vary between and within tumor types, and that only ~ 17% of cancer patients have high OX40 and low OX40L, one of the expression patterns that might be theoretically amenable to OX40 agonist enhancement. Taken together, the data suggest that the OX40/OX40L machinery is a critical part of the immune stimulatory system and that understanding endogenous expression patterns of these molecules and co-existing checkpoints merits further investigation in the context of a precision immunotherapy strategy for cancer therapy.

Phase II basket trial of Dual Anti-CTLA-4 and Anti-PD-1 blockade in Rare Tumors (DART) SWOG S1609: adrenocortical carcinoma cohort

OBJECTIVES

Multiple common cancers benefit from immunotherapy; however, less is known about efficacy in rare tumors. We report the results of the adrenocortical carcinoma cohort of NCI/SWOG S1609 Dual Anti-CTLA-4 and Anti-PD-1 blockade in Rare Tumors.

DESIGN/SETTING

A prospective, phase 2 clinical trial of ipilimumab plus nivolumab was conducted by the SWOG Early Therapeutics and Rare Cancers Committee for multiple rare tumor cohorts across >1,000 National Clinical Trial Network sites.

PARTICIPANTS

21 eligible patients were registered. Median age was 53 years (range 26-69); 16 (76%) were women.

INTERVENTIONS

Ipilimumab 1 mg/kg intravenously every 6 weeks with nivolumab 240 mg intravenously every 2 weeks was administered until disease progression, symptomatic deterioration, treatment delay for any reason >56 days, unacceptable or immune-related toxicity with inability to decrease prednisone to <10 mg daily, or per patient request.

MAIN OUTCOME MEASURES

The primary endpoint was the overall response rate (ORR) (RECIST V.1.1). Secondary endpoints include clinical benefit rate (CBR) (includes stable disease (SD)>6 months), progression-free survival (PFS), overall survival (OS), and toxicity. Immune-related outcomes included immune ORR (iORR), immune CBR (iCBR), and immune PFS (iPFS). A two-stage design was used assuming: null=5% alternative=30%, n=6 in the first stage, 16 max, one-sided alpha=13%.

RESULTS

The median number of prior therapy lines was 2 (range: 1-9). 3 of 21 patients attained confirmed partial response (PR) (ORR=14%). In addition, one patient had an unconfirmed PR; one, stable disease (SD)>6 months; one, immune-related RECIST (iRECIST) PR (iPR); and one patient attained iSD>6 months: clinical benefit rate (response or SD>6 months)=5/21 (24%), iORR=4/21 (19%), iCBR=7/21 (33%). The 6-month PFS was 24%; 6-month iPFS, 33%. The PFS for patients (N=7) with iRECIST clinical benefit were 57, 52, 18, 15, 13, 7, and 7 months. The 6-month OS was 76%; the median OS, was 15.8 months. The most common toxicities were fatigue (62%) and rash (38%), and the most common grade 3/4 immune-related adverse events were hepatic dysfunction (9.5%) and adrenal insufficiency (9.5%). Treatment-related adverse events leading to discontinuation of therapy in four patients (21%). There were no grade 5 adverse events.

CONCLUSIONS

Ipilimumab plus nivolumab is active in refractory metastatic adrenocortical cancer meeting the primary endpoint of the study, with a 19% iORR and 33% iCBR (includes SD/iSD>6 months) and with the longest PFS/iPFS of 52 and 57 months.

TRIAL REGISTRATION NUMBER

NCT02834013 (registered 15 July, 2016; https://clinicaltrials.gov/ct2/show/NCT02834013).

TIM-3 transcriptomic landscape with clinical and immunomic correlates in cancer

TIM-3, an inhibitory checkpoint receptor, may invoke anti-PD-1/anti-PD-L1 immune checkpoint inhibitor (ICI) resistance. The predictive impact of TIM-3 RNA expression in various advanced solid tumors among patients treated with ICIs is yet to be determined, and their prognostic significance also remains unexplored. We investigated TIM-3 transcriptomic expression and clinical outcomes. We examined TIM-3 RNA expression data through the OmniSeq database. TIM-3 transcriptomic patterns were calibrated against a reference population (735 tumors), adjusted to internal housekeeping genes, and calculated as percentiles. Overall, 514 patients (31 cancer types; 489 patients with advanced/metastatic disease and clinical annotation) were assessed. Ninety tumors (17.5% of 514) had high (≥75th percentile RNA rank) TIM-3 expression. Pancreatic cancer had the greatest proportion of TIM-3 high expressors (36% of 55 patients). Still, there was variability within cancer types with, for instance, 12.7% of pancreatic cancers harboring low TIM-3 (<25th percentile) levels. High TIM-3 expression independently and significantly correlated with high PD-L2 RNA expression (odds ratio (OR) 9.63, 95% confidence interval (CI) 4.91-19.4, P<0.001) and high VISTA RNA expression (OR 2.71, 95% CI 1.43-5.13, P=0.002), all in multivariate analysis. High TIM-3 RNA did not correlate with overall survival (OS) from time of metastatic disease in the 272 patients who never received ICIs, suggesting that it is not a prognostic factor. However, high TIM-3 expression predicted longer median OS (but not progression-free survival) in 217 ICI-treated patients (P=0.0033; median OS, 2.84 versus 1.21 years (high versus not-high TIM-3)), albeit not retained in multivariable analysis. In summary, TIM-3 RNA expression was variable between and within malignancies, and high levels associated with high PD-L2 and VISTA checkpoints and with pancreatic cancer. Individual tumor immunomic assessment and co-targeting co-expressed checkpoints merits exploration in prospective trials as part of a precision immunotherapy strategy.

The Role of Total-Body PET in Drug Development and Evaluation: Status and Outlook

Total-body PET, an emerging technique, enables high-quality simultaneous total-body dynamic PET acquisition and accurate kinetic analysis. It has the potential to facilitate the study of multiple tracers while minimizing radiation dose and improving tracer-specific imaging. This advancement holds promise for enhancing the development and clinical evaluation of drugs, particularly radiopharmaceuticals. Multiple clinical trials are using a total-body PET scanner to explore existing and innovative radiopharmaceuticals. However, challenges persist, along with the opportunities, with regard to the use of total-body PET in drug development and evaluation. Specifically, considerations relate to the role of total-body PET in clinical pharmacologic evaluations and its integration into the theranostic paradigm. In this review, state-of-the-art total-body PET and its potential roles in pharmaceutical research are explored.

Tumor-agnostic baskets to N-of-1 platform trials and real-world data: Transforming precision oncology clinical trial design

Choosing the right drug(s) for the right patient via advanced genomic sequencing and multi-omic interrogation is the sine qua non of precision cancer medicine. Traditional cancer clinical trial designs follow well-defined protocols to evaluate the efficacy of new therapies in patient groups, usually identified by their histology/tissue of origin of their malignancy. In contrast, precision medicine seeks to optimize benefit in individual patients, i.e., to define who benefits rather than determine whether the overall group benefits. Since cancer is a disease driven by molecular alterations, innovative trial designs, including biomarker-defined tumor-agnostic basket trials, are driving ground-breaking regulatory approvals and deployment of gene- and immune-targeted drugs. Molecular interrogation further reveals the disruptive reality that advanced cancers are extraordinarily complex and individually distinct. Therefore, optimized treatment often requires drug combinations and N-of-1 customization, addressed by a new generation of N-of-1 trials. Real-world data and structured master registry trials are also providing massive datasets that are further fueling a transformation in oncology. Finally, machine learning is facilitating rapid discovery, and it is plausible that high-throughput computing, in silico modeling, and 3-dimensional printing may be exploitable in the near future to discover and design customized drugs in real time.

Discussion on "Bayesian meta-analysis of penetrance for cancer risk" by Thanthirige Lakshika M. Ruberu, Danielle Braun, Giovanni Parmigiani, and Swati Biswas

Ruberu et al. (2023) introduce an elegant approach to fit a complicated meta-analysis problem with diverse reporting modalities into the framework of hierarchical Bayesian inference. We discuss issues related to some of the involved parametric model assumptions.

Plasma cell-free RNA PD-L1 or tissue PD-L1 protein expression and outcomes with first-line immunotherapy in metastatic non-small cell lung cancer

Background

Tissue programmed death ligand-1 (PD-L1) protein expression is associated with immune checkpoint inhibitor (ICI) treatment benefit in metastatic non-small cell lung cancer (NSCLC). However, tissue PD-L1 protein testing is limited by sampling tumor heterogeneity and fraught with tissue acquisition difficulties. A liquid biopsy-based PD-L1 assay could overcome these limitations of tissue PD-L1 testing.

Methods

An observational cohort of patients with metastatic NSCLC treated with first-line ICI-based therapies were retrospectively assessed for a pre-planned endpoint of median and 3-year landmark overall survival (OS) based upon plasma cell-free RNA (cfRNA) PD-L1 expression by a commercial exosome-free real-time qPCR assay or tissue PD-L1 protein expression (Dako 22C3) performed in CLIA/CAP accredited laboratories.

Results

53 contemporaneous patients in 3 patient cohorts were compared with a median follow-up 34 months. 16 patients were cfRNA plasma PD-L1 positive, including 6 (37 %) tissue PD-L1 negative or tissue quantity not sufficient for testing; 16 were plasma PD-L1 negative but tissue PD-L1 positive; 21 were both plasma and tissue PD-L1 negative. OS was identical whether positive plasma cfRNA PD-L1 expression or positive tissue PD-L1 protein expression (median OS 15 months; 3-year landmark OS 30 %; hazard ratio (HR) 0.97; 95 % CI, 0.44-2.10; p-value = 0.93). Within the positive plasma PD-L1 cohort there was no differing OS whether tissue PD-L1 positive, negative, or unknown (median OS 15 months; 3-year landmark OS 30 %; HR 1.15; 95 % CI, 0.322-4.05; p-value = 0.81). Positive plasma cfRNA expression was associated with a numerically longer median and higher 3-year OS compared to patients lacking PD-L1 expression (median 15 months versus 8 months; 3-year landmark OS 30 % versus 15 %; HR 0.57, 95 % CI 0.26-1.20; p-value = 0.11).

Conclusions

In this retrospective study of real-world metastatic NSCLC patients, plasma cfRNA PD-L1 expression was similarly predictive of ICI benefit as tissue PD-L1 protein expression.

High CTLA-4 transcriptomic expression correlates with high expression of other checkpoints and with immunotherapy outcome

Background

CTLA-4 impedes the immune system's antitumor response. There are two Food and Drug Administration-approved anti-CTLA-4 agents - ipilimumab and tremelimumab - both used together with anti-PD-1/PD-L1 agents.

Objective

To assess the prognostic implications and immunologic correlates of high CTLA-4 in tumors of patients on immunotherapy and those on non-immunotherapy treatments.

Design/methods

We evaluated RNA expression levels in a clinical-grade laboratory and clinical correlates of CTLA-4 and other immune checkpoints in 514 tumors, including 489 patients with advanced/metastatic cancers and full outcome annotation. A reference population (735 tumors; 35 histologies) was used to normalize and rank transcript abundance (0-100 percentile) to internal housekeeping gene profiles.

Results

The most common tumor types were colorectal (140/514, 27%), pancreatic (55/514, 11%), breast (49/514, 10%), and ovarian cancers (43/514, 8%). Overall, 87 of 514 tumors (16.9%) had high CTLA-4 transcript expression (⩾75th percentile rank). Cancers with the largest proportion of high CTLA-4 transcripts were cervical cancer (80% of patients), small intestine cancer (33.3%), and melanoma (33.3%). High CTLA-4 RNA independently/significantly correlated with high PD-1, PD- L2, and LAG3 RNA levels (and with high PD-L1 in univariate analysis). High CTLA-4 RNA expression was not correlated with survival from the time of metastatic disease [N = 272 patients who never received immune checkpoint inhibitors (ICIs)]. However, in 217 patients treated with ICIs (mostly anti-PD-1/anti-PD- L1), progression-free survival (PFS) and overall survival (OS) were significantly longer among patients with high versus non-high CTLA-4 expression [hazard ratio, 95% confidence interval: 0.6 (0.4-0.9) p = 0.008; and 0.5 (0.3-0.8) p = 0.002, respectively]; results were unchanged when 18 patients who received anti-CTLA-4 were omitted. Patients whose tumors had high CTLA-4 and high PD-L1 did best; those with high PD-L1 but non-high CTLA-4 and/or other expression patterns had poorer outcomes for PFS (p = 0.004) and OS (p = 0.009) after immunotherapy.

Conclusion

High CTLA-4, especially when combined with high PD-L1 transcript expression, was a significant positive predictive biomarker for better outcomes (PFS and OS) in patients on immunotherapy.