Durable response rate as an endpoint in cancer immunotherapy: insights from oncolytic virus clinical trials

A growing body of clinical evidence indicates that traditional response criteria may not be sufficient to completely characterize the full spectrum of activities of current anticancer immunotherapies and/or define the clinical benefit associated with their use. Although a number of immunotherapies have recently been approved based, in part, upon tumor responses evaluated by traditional response criteria (eg, Response Evaluation Criteria in Solid Tumors [RECIST] [1, 2] or World Health Organization [WHO] criteria for response [3]), these criteria may not accurately capture the full benefits of immunotherapy-based drugs or regimens. In particular, these measures of response lack a durability dimension, and thus therapeutic benefit may be overestimated. Conversely, because responses may occur after an initial increase in tumor burden or appearance of new tumor lesions characterized as progressive disease by WHO criteria or RECIST [4–11], therapeutic benefit may be underestimated. Consequently, allowing for disease progression before response is critical for evaluation of immunotherapies because initial tumor growth may not always be indicative of treatment failure [4, 11].

Assuming response criteria that allow for progression prior to response are used, a conventional analysis is to evaluate duration of response for responders; however, no prospective criteria are designated for interpreting the clinical relevance of this information at the patient level or for treatment effect in a controlled study. To this end, the endpoint of durable response rate (DRR) has been proposed, which includes standard response criteria and a prospective durability dimension. Although DRR has not been validated as a surrogate endpoint for clinical benefit, anticancer agents have been approved based on a demonstrated ability to achieve retrospectively defined durable responses (DRs) [12–14]. With the allowance for progression prior to response, we hypothesized that DRR is an appropriate measure of the ability of immunotherapy anticancer agents to mediate a clinically meaningful, sustained reduction in tumor burden that may be associated with measures of clinical benefit such as prolonged overall survival (OS), improved quality of life (QoL), and freedom from successive anticancer therapies. This is especially important for immunotherapy-based regimens since the kinetics of response may be different when compared to cytotoxic chemotherapy or targeted therapy that mediate antitumor activity directly in tumor cells.

For this analysis, we used prospectively collected clinical data from the OPTiM trial, a randomized phase 3 study that compared intralesional talimogene laherparepvec, a genetically modified herpes simplex virus (HSV) type-1, designed to selectively replicate in tumors and produce human granulocyte macrophage-colony stimulating factor (GM-CSF) to subcutaneous GM-CSF, to evaluate whether DRR is associated with measures of clinical benefit, such as OS and QoL or treatment-free interval (TFI) after therapy. Talimogene laherparepvec is thought to mediate antitumor activity, at least in part, through activation of systemic immune responses and initial results from previous clinical trials demonstrated delayed kinetics of response similar to observations from T-cell checkpoint inhibitor trials [8, 15]. DRR, defined as a continuous response (complete response [CR]/partial response [PR]) beginning in the first 12 months of treatment and lasting 6 months or longer, was the primary endpoint in OPTiM. Although results from the OPTiM study have been reported previously [15], this analysis was undertaken to further validate the DRR endpoint by determining if DR is associated with other measures of clinical benefit. These data have implications for incorporating DRR as a regulatory endpoint in tumor immunotherapy clinical trial designs.

Data from this analysis of outcomes in the phase 3 OPTiM study indicate that achievement of a DR in patients with unresectable regionally and distantly advanced melanoma was associated with several independent clinical benefits, providing further support for DR as a useful endpoint in patients with melanoma. Achievement of a DR was associated with improved OS in landmark analyses at 9, 12, and 18 months, lower risk of subsequent systemic therapy use at 36 months, and improved QoL at ≥9 months of follow-up. The risk of death was decreased by approximately 90% for patients with a DR. Similar results were obtained in analyses adjusted for imbalances in baseline demographic/clinical characteristics and in a Cox proportional hazards model that evaluated achievement of DR as a time-dependent covariate. Furthermore, patients with a DR have an approximately two-thirds reduction in the risk of requiring a new therapy and were approximately 3 times more likely to have a clinically meaningful improvement in QoL. The association with improvement in QoL is particularly notable in the melanoma setting where skin lesions may be regarded as disfiguring by patients. Taken together, this evidence indicates that DR is an efficacy endpoint that is associated with other favorable clinical outcomes.

Since the proposed mechanism of action of talimogene laherparepvec includes release of tumor-associated antigens and cell- and damage-associated molecular pattern molecules, and local expression of GM-CSF, talimogene laherparepvec is likely able to recruit and expand tumor associated antigen-reactive T cells [21]. This process may take time, and lesions may continue to grow before regression. Furthermore, this process may cause a local inflammatory-like response in lesions, resulting in a period of “pseudo-progression” before response [15, 22]; thus, alternative measures of clinical response may be needed. These observations of progression before response are consistent with the immune-mediated antitumor activity associated with the talimogene laherparepvec mechanism of action, and have been reported with other immunotherapy agents such as ipilimumab [23]. With the allowance for progression before response, our results demonstrate the utility of DR as an appropriate endpoint for oncolytic virus clinical trials and for trials involving combination regimens of immunotherapeutic agents. An element of DR that may be a key to its clinical value is that it incorporates a time component: responses must be maintained for ≥6 months continuously and begun within the first 12 months of initiating treatment. In contrast, the WHO criteria and RECIST do not make allowance for progression before response or define a meaningful minimum duration of response. In a randomized clinical trial of ipilimumab in melanoma, a significant improvement in OS was seen; however, treatment had no impact on progression-free survival (PFS) [10]. These results highlight the inherent problems in using PFS using conventional response criteria as a primary endpoint in immunotherapy clinical trials, and the need for identification of alternative endpoints to accurately assess immunotherapy drugs in early phase clinical development where OS may not be a feasible endpoint. Our data support the use of DRR as a flexible and appropriate surrogate of OS and other measures of clinical benefit.

Other attempts to define alternative clinical trial endpoints for immunotherapy agents include modifications of RECIST (such as immune-related response criteria; irRC [4]), which are being used to evaluate tumor responses in clinical trials across a variety of tumor types [24]. A benefit to the use of irRC is that progressive disease prior to treatment response is permitted, consistent with the criteria for DR in the OPTiM study. However, like the WHO criteria and RECIST, irRC does not define a meaningful duration for tumor response. Furthermore, endpoints such as objective response or progression-free survival based on irRC have not yet been validated as correlates of OS.

Unlike survival endpoints, response indicates a direct biological effect of a therapy on the tumor. Patients can survive for prolonged periods without treatment, but tumors rarely recede or disappear without intervention. To be considered a DR, however, the response must be maintained for a period of at least 6 months. This duration component may make achieving a DR a more descriptive endpoint than others such as objective response, which requires only response confirmation at least 4 weeks after the initial assessment of response. Six months was initially selected as a reasonable duration because most non-immunotherapy treatments rarely induce responses beyond 6 months in melanoma [25]. The expanded use of DRR for other cancers allows for the selection of the most appropriate duration based on the natural history of the disease and known impact of established therapeutic regimens. Thus, by altering the duration of response, DRR can be used in a more flexible manner for other cancers that may exhibit different natural history and/or have other standard treatments available that extend survival for variable times. The inclusion of the initiation at any time within 12 months of starting treatment allows capturing pseudoprogression, which has been reported in some immunotherapy trials, and permits continued treatment beyond clinically asymptomatic progression, as has been reported for irRC proposed for tumor immunotherapy studies [4]. It is also possible that a rate of durable stable disease might represent an informative endpoint; however, further evaluation would be required to validate such an endpoint.

This study had certain limitations. First, these analyses were exploratory and should therefore be interpreted with caution. Second, although we identified associations between DR and clinical outcomes, causative relationships could not be evaluated. For example, although our analysis indicated an association between DR and OS, it is possible that factors other than achievement of DR (eg, imbalances in disease stage and line of therapy, or subsequent therapy) might have influenced the observed outcome. Although endeavors were made to mitigate the potential for confounding by such factors, it is difficult to determine how successful they might have been. Additionally, the selection of 6 months as the minimum duration for achievement of a durable response was based on responses to cytotoxic chemotherapy agents; this time point was, to an extent, arbitrary. In the OPTiM study, some patients had DRs that lasted substantially longer than 6 months. It is unknown whether the associations found in this analysis would have been observed had the distribution of DR duration been closer to the 6-month minimum. Finally, it is possible that the associations seen were driven by the preferential occurrence of complete response among patients with DR (of any duration).

Results from this analysis indicate that the achievement of a DR in patients with unresectable regionally and distantly advanced melanoma was associated with improved clinical benefit and QoL. Although causal relationships cannot be determined via these analyses, these findings support the utility of DR as a meaningful trial endpoint when assessing efficacy of immunotherapies for solid tumors. Further consideration of durable response rate as a clinical endpoint in cancer immunotherapy trials is also warranted and may allow for improved selection of promising agents for later phase clinical development.