Center for Global Research Data

Early markers of clinical outcome to ipilimumab therapy for advanced melanoma

Lead Investigator: Michael J. Sorich, Flinders University of South Australia
Title of Proposal Research: Early markers of clinical outcome to ipilimumab therapy for advanced melanoma
Vivli Data Request: 7140
Funding Source: None
Potential Conflicts of Interest: None

Summary of the Proposed Research:

Immune checkpoint inhibitors (ICIs) represent a very important recent development in the treatment of advanced melanoma. Notably, a subset of patients responds very well to ipilimumab, a types ofICI treatment, and this subset have significantly improved survival. The ability to predict in advance the individuals who are unlikely to benefit is likely to be valuable for guiding decisions regarding the duration and choice of therapy. A number of small studies suggest that early changes in absolute lymphocyte count (ALC), absolute eosinophil count (AEC) and serum lactate dehydrogenase (LDH) may predict response and survival for individuals using ipilimumab (Ku 2010, Delyon 2013, Simeone 2014). However, the smalI sample size and differences in methods and reporting limit the ability to evaluate the validity and comparative performance of these markers. These are also very preliminary evidence that LDH and ALC changes may also be of value for a different type of ICI therapy – PDI inhibitors (Martens 2016, Diem 2016). A small study has also suggesting that changes in AEC may predict risk of immune related adverse events (Schindler 2014). A larger and more comprehensive evaluation of the validity and nature of these associations is required before clinical use of such markers may be realistically considered.

The research proposed will utilise data from studies of ipilimumab therapy for advanced melanoma to evaluate whether early (week 6) changes in selected putative laboratory markers are associated with subsequent response, toxicity and survival. The association will be primarily evaluated by Cox proportional hazards regression for survival outcomes and logistic regression for best overall response and toxicity. Results will be communicated via publication in a peer-reviewed international cancer scientific journal and presented at scientific conferences.

Study Design: Retrospective secondary cohort analysis of relevant participants in existing clinical studies.

Statistical Analysis Plan:

Changes in three putative laboratory markers{ALC, AEC, and LDH) at week 6 compared to pre-treatment will be evaluated for association with overall survival, best response and immune -related adverse events. Pre-treatment levels are defined as the value closest to and before commencement of treatment (max of 14 days prior to commencement of treatment). Week 6 value will correspond to the value associated with the scheduled week 6 visit following commencement of therapy. If this value is missing, the value of the laboratory marker closest to week 6 (within the range of week 4 to week 8) will be used. Week 6 was chosen on the basis that labs were planned at the time point across most studies and it represents both a relatively early time point following commencement of therapy and sufficient time for changes in the laboratory markers to take effect.

The association with OS and response/toxicity will be modelled using Cox proportional hazard regression and logistic regression respectively, stratified by study. Only individuals alive and on-therapy at the landmark time will be included in the analyses undertaken. Additionally, for the evaluation of immune-related toxicity, individuals experiencing the toxicity outcome prior to the landmark time will be excluded from the analysis.

For comparison with prior studies and simplicity, the change in laboratory value at week 6 (vs pre-treatment) will be initially modelled as a dichotomised variable (increased vs decreased value at week 6 compared to pre-treatment). The association will be reported as a hazard (odds) ratio with a 95% confidence interval and p value (Wald test), and graphically via a Kaplan Meier plot for survival outcomes. To complement this dichotomised version of the explanatory variables, the association between the change in the laboratory value as a continuous variable and the relevant outcome will be reported graphically based on modelling using restricted cubic splines.

A range of exploratory and sensitivity analyses will be undertaken to provide further insight regarding the results of the primary analysis. All exploratory analyses will be explicitly labelled as exploratory in reporting of the study.
Exploratory and sensitivity analyses will evaluate:
l. whether information provided by changes in the laboratory predictors provides any additional information over what can be assessed pre-treatment. This will be undertaken using multivariable analyses including established and plausible pre-treatment variables. For the survival and response outcomes this will include visceral involvement, serum LDH, relative lymphocyte count, relative eosinophil count , number and type of previous therapies, baseline tumor size, dose schedule, sex, age, ECOG status, BRAF mutation status, M stage, number and location of metastases.

2. whether results are sensitive to the time point assessed (time points ≤ 12 weeks, with particular interest in exploring the earliest time-point that changes in the variables are sufficiently predictive of outcomes), the incorporation of information from multiple timepoints (slope of change or average over first 6 weeks), and the choice of dichotomization cut point.

3. whether changes in laboratory predictors depend on the respective baseline (e.g. prior reports suggest that LDH changes may only be predictive for individuals with high baseline LDH (Diem 2016)).

4. whether the putative marker s are equally useful for predicting each type of tumor response to ipilimumab therapy, in particular delayed response.

5. whether change in AEC is equally useful for predicting each type of immune-related adverse event, and early vs late adverse events.

6. whether relationships identified are specific to ipilimumab therapy. The association of the putative laboratory markers with best response and survival will also be modelled for non-ICI (e.g. chemotherapy) comparator arms of the clinical studies available.

Requested Studies:

MDX-010 Antibody, MDX-1379 Melanoma Vaccine, or MDX-010/MDX-1379 Combination Treatment for Patients With Unresectable or Metastatic Melanoma
Data Contributor: Bristol Myers Squibb
Study ID: NCT00094653
Sponsor ID: CA184-002

Phase II Study to Determine Predictive Markers of Response to BMS-734016 (MDX-010)
Data Contributor: Bristol Myers Squibb
Study ID: NCT00261365
Sponsor ID: CA184-004

A Study of MDX-010 (BMS-734016) Administered With or Without Prophylactic Oral Budesonide
Data Contributor: Bristol Myers Squibb
Study ID: NCT00135408
Sponsor ID: CA184-007

A Single Arm Study of Ipilimumab Monotherapy in Patients With Previously Treated Unresectable Stage III or IV Melanoma
Data Contributor: Bristol Myers Squibb
Study ID: NCT00289627
Sponsor ID: CA184-008

Monoclonal Antibody With or Without gp100 Peptides Plus Montanide ISA-51 in Treating Patients With Stage IV Melanoma
Data Contributor: Bristol Myers Squibb
Study ID: NCT00077532
Sponsor ID: CA184-021

Study of Ipilimumab (MDX-010) Monotherapy in Patients With Previously Treated Unresectable Stage III or IV Melanoma
Data Contributor: Bristol Myers Squibb
Study ID: NCT00289640
Sponsor ID: CA184-022

Dacarbazine and Ipilimumab vs. Dacarbazine With Placebo in Untreated Unresectable Stage III or IV Melanoma
Data Contributor: Bristol Myers Squibb
Study ID: NCT00324155
Sponsor ID: CA184-024

Evaluation of Tumor Response to Ipilimumab in the Treatment of Melanoma With Brain Metastases
Data Contributor: Bristol Myers Squibb
Study ID: NCT00623766
Sponsor ID: CA184-042

Summary of Results:

This project was submitted to the trial sponsor in 2016 and due major limitations of the original (non-Vivli) research environment, the analyses required were not able to be undertaken. We are grateful that the trial sponsor was able to make the data available via Vivli in 2021. Unfortunately, the fast changing nature of the field of immune-oncology meant that the original specific research questions defined in 2016 were no longer of major contemporary clinical interest.