Cancer is basically a process of the patient’s own cells dividing rapidly and failing to die normally. For a cancer to become established in a host, the transformed cells must also develop mechanisms to avoid eradication by the immune system. Therapeutic manipulation can activate the innate immune system leading to cell death and, under appropriate conditions, innate and adaptive immunity leading to oncolysis while promoting long-term memory responses.

I-O therapy involves a fundamentally different approach from conventional chemotherapy, which unleashes an indiscriminate, static, and toxic direct attack on all cells – malignant and normal -- in hopes of damaging the cancer cells more than the host cells. Recent studies have suggested that cytotoxic chemotherapy and targeted therapy may also target stromal cells and immune cells within the tumor microenvironment [7, 8]. These observation suggest the potential for combining chemotherapy with IO agents with the goal not of killing as many tumor cells as possible, but rather to optimize immunologic clearance, which may allow for lower chemotherapy dosing. Because immunotherapy for cancer primarily relies on an indirect approach rather than a direct attack on cancer cells, the observed kinetics of response related to I-O therapies can be delayed [9] and, at times, the tumor may appear to be growing in the near term, when in fact the observed increase in volume is instead related to an inflammatory immune response that is working to eliminate the cancer [4].

An adaptive immune response is characterized by the ability to persist, creating “immune memory” that, once effectively triggered by an immunotherapy, can enable the body to maintain an ongoing defense against a threat like a virus or a cancer cell expressing specific antigens, even after therapy is discontinued and perhaps for the lifetime of the patient. I-O therapies can also evolve over time, broadening and deepening anti-tumor immunity, preventing the cancer’s ability to escape through the selective growth of variants that can evade immune detection. The rapid co-evolution of tumor cells and immune responses may also result in immunoediting resulting in loss of antigen-specific immunity explaining, in part, IO drug resistance and the need to reconsider the pharmacologic drug class, dosing, schedule and combination to optimize anti-tumor activity [10].

Evidence of an effective and durable immune response against cancer dates back more than three decades, as high-dose interleukin 2 (IL-2) therapy produced durable responses with few relapses among approximately 10% of patients with advanced renal cell carcinoma (RCC) and melanoma [11]. These experiences demonstrated a unique hallmark of immunotherapy for the treatment of cancer: the flattening of the Kaplan Meier survival curve, in which a long, plateau of the curve represents durable responses that, for some patients, may extend throughout their lives.

With the advent of checkpoint inhibitors as single agents and in combination, dramatic results were first seen in patients with melanoma. The proportion of patients with metastatic melanoma experiencing objective responses increased to 20–22% with ipilimumab (anti-CTLA 4) treatment and 35–40% with anti-PD-1 agents, and above 50% with a combination approach [1].

Similarly, significant results with checkpoint inhibition approaches have yielded regulatory approvals of novel drugs and combination regimens, leading to new standards of care for patients with RCC, non-small cell lung cancer (NSCLC) [12], small cell lung cancer (SCLC) [13], bladder cancer [14], Merkel cell cancer [15], head and neck cancer [16], gastrointestinal cancer [17]and certain lymphomas [18]. Investigators are motivated by early success in identifying potential predictive biomarkers to select patients most likely to benefit (including programmed death ligand-1 or PDL1, and micro-satellite instability high or MSI-Hi), as checkpoint inhibition strategies are yielding even higher response rates in some tumors [19, 20]. Durable responses have also led to FDA approval of two CAR-T cell approaches for the treatment of acute lymphoblastic leukemia in children and young adults, and in certain forms of non-Hodgkin lymphoma in adults [21, 22].

In general, immune checkpoint inhibitors have been associated with immune-related adverse events while CAR T cell treatment has been associated with cytokine release syndrome and neurologic toxicities. While serious adverse events are rare, mortality has been reported for patients receiving immune checkpoint inhibitors [23]. Nevertheless, I-O treatment has been suggested to have less impact on patients’ quality of life than conventional therapies [24], especially when adverse events are expeditiously managed early with corticosteroids and other immunosuppressive agents [25]. This parallels the experience with CAR-T research, in which cytokine release syndrome (CRS) was identified as an early potentially lethal clinical syndrome [26, 27], but an effective clinical management strategy was quickly identified [26], which did not appear to interfere with efficacy [28, 29], and actually led to a concomitant FDA-approved indication expansion for the anti-IL-6 monoclonal antibody, tocilizumab, since IL-6 is believed to be a major cytokine released in patients experiencing IO-induced cytokine release syndrome [30]. Research is ongoing to better define the most serious immune-related adverse events and identify patient characteristics most likely associated with them (recognizing that patient cohorts in most pre-approval studies did not fully reflect the general population) [31, 32].

Because I-O therapies act on the immune system, they may be effective if administered for a shorter period. As a result, many I-O treated patients experience significant “Treatment-free survival” (TFS), the period that occurs after treatment ends, and while the impact of the therapy endures, patients may not require other treatment(s) [33]. TFS provides an important opportunity for patients and their families to resume routine activities, travel, and generally approach their daily lives free from ongoing cancer treatment [34].

There may also be financial benefits to individual patients, their families, and society that result from patients being able to return to work earlier and for longer periods of time while also reducing the need for additional or subsequent cancer treatments and perhaps less frequent medical tests and interventions. When effective, I-O treatment should boost productivity for many patients and may save individuals, families, and society considerable expenditures throughout the rest of their lives.

While there is risk of serious toxicities associated with current I-O regimens, I-O therapies generally do not lead to the side effects commonly associated with cytotoxic chemotherapy such as nausea/vomiting, hair loss, and risks to fertility. In fact, the knowledge about 1) what side effects are likely to occur from I-O therapies and 2) that most can be managed in the near-term (by experiences providers) without impacting the effect of the cancer treatment, adds to patients’ current willingness to try them – especially when faced with few other potentially curative treatment options.

Late-stage patients facing the possibility of dying from their cancer often value the opportunity to pursue a hopeful gamble and receive a novel therapy that offers the potential for long-term disease control in a small percentage of patients rather than a treatment that offers potential benefit to a higher proportion of patients but for a shorter duration. Further, patients often will place a higher value overall on survival than their clinicians, who typically focus more on progression-free survival and managing patient’s treatment and disease related symptoms [35]. Of course, there are other factors at play in determining whether a patient has access to such hopeful gamble therapies, e.g. geographic access to healthcare provider expertise in IO delivery, drug availability, negative reimbursement incentives, high out-of-pocket expenses and others, raising important issues for society that are beyond the scope of this paper.

Reports of significant positive outcomes with I-O therapy for an increasing number of tumor types have fueled hope among patients for long-term survivorship and even cure in some cancers. This type of hope – especially for patients with dismal prognoses -- has been recognized to provide positive benefits to the patient’s quality of life [36] and is a powerful incentive for patients to seek access to these therapies, even while recognizing the longer odds of success. There is active debate within the oncology community about if and when to try immunotherapies when patients have few other valid options, even though the evidence is not yet conclusive about the potential benefit [37]. This may be especially important for patients with orphan cancers where clinical trials are lacking and where few approved agents are available.

Traditional clinical outcome measures, or clinical outcome assessments (COAs), in trials include overall survival (OS), progression-free survival (PFS) and objective response rate (ORR). These have long proved to be useful measures for assessment of cytotoxic chemotherapy, but a more complete assessment of the value of I-O requires identifying and measuring the impact of I-O therapy on patient’s lives. Some I-O therapy studies have shown significant improvement in overall survival without any impact on PFS, making the use of OS surrogates problematic in value frameworks that are not accounting for the potential differences in endpoint analyses.

A recent review by the ISPOR (International Society for Pharmacoeconomics and Outcomes Research) Special Task Force on US Value Frameworks has identified multiple value frameworks in the U.S. [41] In Europe, where HTA bodies are much more prevalent, there is less need for discrete value frameworks, but the European Society for Medical Oncology (ESMO) has created one based on “magnitude of clinical benefit.” [42] Others strive to be more patient-centered, emphasizing the patient experience [43]. In addition to understanding how each framework defines 'value', it is also important to consider that those designed by clinically-oriented bodies are meant to inform clinician-patient decisions, while those geared for payers are meant to inform payer and pharmacy benefit manager decision-making around coverage or formulary tiering.

The report of the Second Panel on Cost-Effectiveness in Health and Medicine (Second Panel) has defined four normative perspectives for consideration in evaluating value: 1) the payer perspective; 2) the health care sector perspective; 3) the health care sector with time cost perspective; and 4) societal perspective [44]. While each is scientifically valid and informative for their respective decision makers, the Second Panel recommended that analyses should include “reference cases” from the health care sector perspective and the societal perspective, which could be helpful in understanding how the value assessment informs a comparison within the therapeutic class or across therapeutic classes. Some stakeholders have noted a shortcoming in the Second Panel’s work, noting that it did not specifically call out patient perspectives in its report [45].

While some observers have criticized the recent value frameworks [46, 47], those meant to inform clinician-patient decisions do have elements of patient preference included in them, which may make the ‘value’ resulting from them reflective of an individualized assessment, and possibly then fit for informing individualized clinician-patient decisions. More payer-centric value frameworks also include elements of patient preferences but given the goal of informing population-level decision making, such value estimates are conducted at the average of a population. Thus, heterogeneity in individual patient preferences are often lost in these population-geared exercises.

Because of the mechanistic differences between I-O therapies and traditional chemotherapy, conventional trial designs and endpoints generally do not fully capture the novel patterns of treatment response. This unique aspect of I-O suggests that longer-term assessment at multiple timepoints is needed to adequately evaluate outcomes [48]. Traditional parametric survival models used commonly to estimate long-term survival cannot adequately represent complex hazard functions and may not be appropriate for modelling the underlying mechanism of action associated with I-O treatments [49].

Recent work reported by the ISPOR Special Task Force in rare pediatric diseases presents some of the unique challenges in selection of clinical outcome assessments (COAs), and highlight the importance of developing uniform methods and metrics to capture relevant outcomes of interest for the I-O setting [50]. Additionally, recent work from the ISPOR Rare Disease Special Interest Group has identified several key challenges to research in rare diseases, which may be particularly relevant for I-O, and result in a lack of tailored health technology methods for rare disease treatments, as well as significant uncertainty for HTA authorities [51]. Many of the factors result from the evolving evidence base, including difficulties in establishing specific and sensitive diagnostic criteria, and evaluating the treatment effect (or heterogeneity of treatment effect). Combined with ethical challenges in designing appropriate clinical trials, insufficient knowledge of the natural history of the disease, and often poor patient recruitment for trials, the result is high levels of uncertainty in assessing value for these therapies. These uncertainties are factored into health technology assessments by global authorities, as comprising the level of certainty that is generally attributed to the value of a product. In addition, the model structure may not reflect the full patient experience, often failing to assess the value of treatment-free survival.

While the long-term clinical and economic impact of safety monitoring with I-O therapy is not defined, current practice suggests that limited baseline screening and on-going laboratory monitoring with detailed clinical surveillance and patient education can identify adverse events early, allowing rapid intervention [52]. Whether this results in better compliance with planned treatment duration or prevents chronic toxicity is unknown. The optimal duration of treatment with I-O has also recently undergone considerable debate and discussion with some clinicians suggesting that early drug discontinuation may be possible without increasing rates of tumor progression [53].

An improved understanding of tumor immunology has led to new combination treatments, although it is unclear whether concurrent or sequential administration impacts outcomes. Further studies will focus on better defining effective combination regimens, treatment schedules and duration of therapy while refining safety monitoring measures that will allow appropriate patient management while limiting unnecessary diagnostic work-ups. These advances in limiting and mitigating toxicities should provide additional I-O relevant evidence to support better value assessments. There is a need for long-term follow up via accurate registries, capturing patient outcomes in community settings as well as academic medical centers.

One of limitations of reliance on the QALY within certain value frameworks is its primary dependence on survival endpoints (or improvements in OS and/or PFS) in determining the incremental cost per QALY gained for interventions that have OS and/or PFS primary endpoints in clinical trials [54]. Indeed, the ISPOR Special Task Force on Value Frameworks echoed the recommendation that cost-effectiveness analysis “as measured by cost per QALY [should serve] as a starting point to inform payer and policy maker deliberations” [55]. A natural question arises as to whether or not the QALY can be a comprehensive estimate of health outcome for the purposes of characterizing I-O therapies. Some cases of incremental cost-per-QALYs for I-O therapies suggest good value for money [56]. However, the question remains as to whether QALYs are sufficiently comprehensive to address the unique long-term outcomes for I-O, especially when compared to more traditional chemotherapy and targeted therapy regimens.

There is increasing interest in ‘going beyond QALYs’, to measure and systematically incorporate patient reported outcomes (PRO) in oncology [57–59], as there are signals (from markets outside the U.S.) that surrogate endpoints like PFS may not be closely associated with improvements in health-related quality of life in oncology clinical trials [60], or that current health-related quality of life instruments lack uniformity when applied across therapeutic areas [61]. While various work has suggested how to set standards for PRO use for cancer clinical trials with international standards [62], or in clinical trial protocols [63], there is more to be done before this work is ready for inclusion in value assessments. In fact, a recent FDA analysis has noted that health-related quality of life components most impacted by anti-PD-1/PD-L1 therapies (including disease symptoms, symptomatic toxicity and physical function) have been ‘variable,’ but that “these data, along with other important clinical data such as hospitalizations, ER visits and supportive care medications can help inform the benefit risk assessment for regulatory purposes.” [64]

In the U.S., the Centers for Medicare and Medicaid Services (CMS) has recently opened a National Coverage Determination (NCD) for Chimeric Antigen Receptor T-cell (CAR-T) Therapy for Cancers [65] and has focused on the PRO instruments themselves, and whether sufficient scientific evidence exists to support application of PROs to health outcomes research [66]. Presentations by the FDA and PRO experts provided optimism for several of the PRO instruments [67], and a final recommendation from the MEDCAC in the form of a proposed Decision Memo is expected in 2019 [68].

There is increasing interest in incorporating more patient centric elements in value assessments, especially as recent evidence appears to suggest an OS improvement among metastatic cancer patients who had PROs integrated into their routine care, compared to usual care [69]. While Basch had previously pointed out the lack of PRO data in existing value frameworks [70], he also argues for greater uniformity in how the PROs are incorporated into the value assessment for CAR-T cell therapies and to include patient representatives in consensus processes. While there seems to be increasing use of validated PRO instruments in oncology clinical trials, there are challenges to incorporating the PRO measures into existing value frameworks [71].

It is also challenging to weigh the different trade-offs between therapies in a class and the added layer of complexity associated with evaluating combination therapies. Likewise, there is the challenge of distinguishing between novel I-O therapies and their chemotherapy comparators, with the concept of treatment-free survival raising additional questions for researchers to address. An emphasis on integrating data collection regarding both PRO and quality of life (QOL) into modern I-O clinical trials will be important to developing benchmark metrics for understanding the impact of these measures related to specific drug agents and tumor types. The development of benchmark data will also provide a basis for comparisons to patient outcome data with more traditional cancer therapeutics.

Post-market research is important to our understanding of the costs and real-world effectiveness of novel therapy approaches post launch. An important aspect of measuring real-world effectiveness is comparison of available treatment options in real world populations (i.e. comparative effectiveness). Thus, careful consideration for study design is needed not only to collect important elements of value but also to ensure that observed signals can be attributable to the I-O therapy.

While evidence to support costs and real-world effectiveness estimates improves, researchers should advance models that support informed decisions. This may include, but is not limited to, increased modeling transparency [74], clearly outlining data and underlying assumptions used for calculations [75, 76], consensus on value elements [77] to incorporate into individual assessments, and continuous patient engagement [78, 79] throughout the process to ensure a patient-centric approach.

We recommend a concerted effort to develop models for looking beyond the median and conducting appropriate pre-planned sub-group analysis of the patients who see long-term benefit (e.g. “the Tail of the Curve” phenomenon, which within oncology, is seen by clinicians and patients as a defining hallmark of I-O). Table 2 describes considerations for such I-O specific elements to enhance a traditional ICER calculation.Tab2

I-O specific elements to enhance traditional value calculations