Advanced APMs and the Emerging Role of Immuno-Oncology Agents: Balancing Innovation and Value
Michael V. Seiden, MD, PhD; Marcus Neubauer, MD; and Diana Verrilli
As the cost of healthcare in the United States crosses $3 trillion and cancer care approaches $150 billion annually, there is a growing public discourse on strategies to mitigate healthcare expenditures.1
Commercial payers, government agencies, and in particular, HHS, have begun piloting value-based reimbursement strategies to see how these perturbations might encourage the evolution of clinical practice and care delivery towards a value-based, alternative payment model (APM). The Center for Medicare and Medicaid Innovation (CMMI) Oncology Care Model (OCM)—proposed, and now active, in 196 predominantly community oncology practices across the country—will be particularly important, both due to the large number of eligible participants, the complexity of the program design,2
and the wealth of data that will be generated and shared between the payer (CMS) and the participating practices. If this program is successful at reducing Medicare program expenditures, while preserving or improving the quality of cancer care for beneficiaries, it likely will have a profound impact on how cancer care is paid for in the future by both government and commercial payers.
APMs that incentivize reductions in the total cost of cancer care (like the OCM) will encourage providers to focus on 2 of the largest categories of spend: avoidable hospital admissions and variations in treatment protocols, with a particular focus on clinical pathways and drug costs. While no drugs are singled out in the OCM model, the rapid increase in the use of immuno-oncology (I-O) agents—programmed death-1 (PD-1) and now programmed death ligand-1 (PD-L1) inhibitors—in oncology suggests that the expanding use of these agents will provide specific challenges to practice success in this program, and will likely present both operational and moral challenges for physicians attempting to succeed in the OCM.3
The Opportunity Offered by I-O in 2016 and Beyond
In recent years, a deeper understanding of how the immune system works, and how cancer disarms typical immune effector functions has led to advances in ways to improve the function of specialized immune cells, called T cells.3
While many strategies are under development, this review will focus on antibodies that bind chimeric T-lymphocyte associated protein 4 (CTLA-4), PD-1, or PD-L1 in such a way that it interferes with the binding of these cell surface proteins with their natural ligands.1
Vaccines, chimeric antigen receptor (CAR)-T cells, and small molecules that interfere with or augment other immune functions will pose similar challenges in the future, but still require further development and will not be discussed further, except to emphasize that in toto the development of I-O agents, either alone or in combination strategies, will likely dominate therapeutic advances in oncology for the next 10 years.
CTLA-4, PD-1, and PD-L1–Targeted Agents
Antibodies that bind CTLA-4 (ipilimumab), PD-1 (nivolumab and pembrolizumab), and PD-L1 (atezolizumab) successfully reactivate an exhausted or deactivated immune response.3
Once reactivated, some tumors will demonstrate an expansion of activated immune cells within the tumor with rapid, cancer cell destruction. Single-arm and randomized studies in melanoma, lung cancer, renal cell carcinoma, head and neck cancer, bladder cancer, and Hodgkin’s disease have demonstrated clinical activity of many of these agents.4-10
While the activity in Hodgkin’s disease and melanoma has been dramatic, and in the case of melanoma, often very durable (measured in years), the findings in solid tumors have demonstrated a few key themes as reviewed in Table 1
What is most intriguing is not the response rate associated with I-O, instead the tail of the survival curve (Figure 1
). In essentially all the studies that used I-O agents, there is a subset of 10% to 20% of patients who demonstrate durable responses with marked clinical improvement persisting well more than a year after treatment. Some studies with longer follow-up have demonstrated multi-year responses.4-11
Durable responses are distinctly unusual in patients treated with standard chemotherapy or molecularly targeted agents that inhibit nonimmune targets.
For patients with advanced cancers being treated with palliative (noncurative) intent, despite the wealth of data suggesting a very guarded prognosis, their primary question is “can my cancer be cured?” or as a compromise “can my cancer be treated so I can live a lot longer?” If I-O provides hope for multi-year and perhaps decade-plus remissions, even if that likelihood is low (say 10%), all patients will want a chance at winning on what has been described as “an I-O lottery.” Further, patient and physician preference for I-O is also influenced by the fact that, compared with many alternative therapies, I-O is less toxic for most patients than chemotherapy. Thus, PD-1 inhibitors gain a strong acceptance based on slightly superior response rates, the important “tail of the curve” comprised of the I-O lottery winners,” and decreased toxicity.
I-O at the Center of the Clash of Innovation and Value
The management of cancer is complex and expensive. The use of sophisticated imaging, molecular diagnostic tests along with the frequent use of expensive therapeutic modalities including surgery, radiation, chemotherapy, biologics, and supportive drugs can easily lead to courses of therapy that exceed $100,000 in the first year of a cancer diagnosis. Similar or even greater expenses are often also seen in the last year of life for those patients who can’t be offered curative therapy. Cancer spending was estimated to be $130 billion in 2010 and is expected to be between $170 billion and $180 billion in 2020.1
demonstrates the rough breakdown of annualized costs from patients with cancer. As seen in the graph, a significant expense is the office-based drug costs. Cancer drugs (oral plus intravenous) and hospitalization make up almost two-thirds of the spend, thus, it will be hard to ignore these categories of expense as the provider community is challenged to participate and ultimately take risk in the quest to decelerate the increasing spend on cancer care. A special area of focus will be the PD-1 and PD-1–inhibitors, which have gone from investigational agents to one of the top 5 oncology drugs in terms of costs in the community, in the last 2 years, with year-over-year growth of 300%. With the recent approval of pembrolizumab in a subset of individuals with newly diagnosed metastatic lung cancer, the country’s most common lethal malignancy, it is likely even more patients will be getting this therapy with their time on therapy being potentially longer. In addition, it is likely that in 2017 we will see additional inhibitors of PD-1 and PD-L1 approved, and potentially the combination of these agents with ipilimumab or other CTLA-4–binding agents. By some estimates, I-O drugs already represent 18% of the drug spend. As the principal APM in cancer, the OCM is due to pilot through 2021, and it is likely that I-O drugs will represent the principal driver of cancer drug costs, which as a class may represent 50% of the oncology drug spend towards the end of the OCM pilot trial.
Alternative Payment Models
I-O innovation arrives concurrently with the rapid roll-out of APMs. Both government and commercial payers are exploring novel, alternative payment strategies that will compel physicians to pay considerably more attention to the total cost of the cancer treatment and evaluation they prescribe. As the average age of cancer incidence is adults in their 60’s, Medicare is the single most dominant payer of cancer expenses in this country. Thus, it is instructive to consider the Medicare Access and CHIP Reauthorization Act (MACRA) and the OCM to illustrate how these programs will impact oncology care and the oncology practitioner.
In 2015, Congress repealed a long-standing CMS program called the Sustainable Growth Rate program and replaced it with MACRA, which includes a performance-based adjustment to a practice’s Medicare payment, under the acronym of MIPS (Merit-based Incentive Payment System). This program, which took effect in January 2017, requires the submission of quality data from the practices along with claims data (from CMS). Practices will be scored on clinical quality metrics, cost of care, and practice improvement. CMS will provide 0.5% annual increases in total payouts through 2019 and will then keep payouts flat (budget neutrality), but will rank practices based on the above metrics to pick winners and losers. Those who outperform their peers will receive bonus payments of up to 9% while those that significantly underperform will see their payments drop by 9%. Of note, a well-run oncology practice in the community might expect a margin of 12% to 18% of revenue; thus a 9% gain or loss will have extraordinary effects on the bottom line of the clinic, especially when one considers that overachieving on quality metrics might take incremental investments. Practices that invest wisely and over-perform can expect at least some financial reward. Those that lack efficiency, scale, or a focus on quality will likely face extreme challenges in the MACRA-MIPS environment. Data will be sent to CMS in 2017, with bonuses or penalties being delivered in 2019.
A practice can gain additional benefits and risks by entering an advanced APM such as the “2-sided risk” version of OCM, which includes some of the same quality metrics of MACRA and MIPS along with additional responsibilities in transforming the care delivered to the patient. Practices participating in the OCM are compared with their baseline performance prior to program initiation, along with a number of adjustments and trend factors. The trend factor, still poorly defined, is proposed to correct for inflation and take into account the realization that innovation will likely increase costs. Of note, the global baseline cancer spending calculated by CMS preceded the approval of all the PD-1 and PD-L1 checkpoint inhibitors.
Defining Value in Healthcare
In healthcare, value is often defined as quality divided by cost—increasing quality and/or decreasing costs offers the chance of increasing value. Indeed, in theory, one could still achieve an increase in value with higher costs if the quality of care improved dramatically. Economists often refer to value from the patient perspective. Measuring cost, the denominator of the value equation, is not as simple as it seems because it includes not just the direct cost of care, but also opportunity costs for the patient from lost time at work, or perhaps for the caregiver who left work to support the patient. Quality is an order of magnitude more challenging to convert into a number. From the patient’s perspective, quality could include several concerns:
Do I feel well?
Can I live independently?
Is my anxiety and/or pain relieved?
Do I feel cared for?
Can I return to work?
Obviously, how an individual patient scores these issues, numerically, is a daunting proposition and likely is very individualized. One patient might emphasize that value centers predominantly around their ability to live independently, while another might ascribe value just to surviving to a key milestone such as a wedding or delivery of a grandchild. It should be apparent that these types of value determinations are qualitative and precise measurement is aspirational. As a poor substitute, organizations have proposed various statistical standards such as cost per quality-adjusted life-years that a certain therapy provides, or costs to prevent a single death.
Within the US Oncology Network—a national collection of 1400 cancer caregivers including over 800 medical and gynecologic oncologists—much of the decisions around drug use are guided by well-conducted and peer-reviewed clinical trials that are vetted by the National Comprehensive Cancer Network, and then further scrutinized by a Pathway Committee of the US Oncology Network. Selected drug regimens are defined by a combination of clinical efficacy and cost. Typically, costs are simply the drug costs since there is little data of costs of supportive care, in particular the likelihood that a specific regimen will require urgent care, emergency care, or hospitalization. Developing such databases personalized to an individual’s age, gender, comorbid disease, and performance status is an important goal for the future.
Ethical Issues of Society Versus the Individual and Models Around the Drug Value Proposition
The I-O agents, specifically, the PD-1 and PD-L1 inhibitors, pose a particular set of challenges. Although sweeping generalizations are hazardous, a high-level summary of the data around these agents is as follows:
Compared with chemotherapy or other targeted agents, I-O agents are marginally more effective in many patients, and moderately to markedly more effective in a small minority of patients.
These agents are, in general, less toxic than most chemotherapy agents used in late- stage cancer.
I-O agents are moderately or markedly more expensive than standard chemotherapy.
There is little guidance in the current literature on how long to continue these agents in patients with durable responses or how much emphasis to place on PD-L1 tumor expression.
Thus the cost calculation of the value equation is high; however, since the toxicity is less (on average) than chemotherapy and the clinical outcome is improved, the quality (numerator) of the value equation is also higher. However, its impact on the overall value depends on how much an individual patient benefits, which of course is hard to predict a priori.
To make issues a bit more complicated, alternative payment models, in general, and the OCM in particular, doesn’t really reward value as described above. It does include quality metrics, but doesn’t include response, survival, or toxicity. The calculation of shared saving that would be delivered to the practice is equal to a quality score multiplied by cost saving score. Consider, as an example, a practice that doubled its quality score from 50 to 100 while its healthcare costs increased 10%. In this hypothetical case there was a marked improvement in quality and perhaps even survival, simultaneously with an increase in costs. While the improvement in quality would exceed the cost increase (and thus improved “value”), the cost-saving equation would yield no financial return for the practice (reward in the OCM = quality score x cost savings). In the event the practice made significant financial investment to transform clinical care through investment in information technology, staff, and processes, they could easily have overspent their monthly management fee. This highlights an important discordance with value and shared savings that dominates APMs.
In reality, while quality improvements are important, the dominant factor in the model is a reduction in costs. This puts the physician in a challenging position, especially in a society with direct-to-consumer marketing and the increasing awareness and hype around the potential of immunotherapy. For a patient with recurrent cancer of the lung, kidney, bladder, head and neck region, Merkel cell tumor, melanoma, or Hodgkin’s disease, there is a growing body of literature that supports considering I-O agents. There are currently hundreds of clinical trials evaluating PD-1 and PD-L1 agents alone or in combination with other I-O agents. It is almost certain that during the tenure of the OCM project, PD-1 and PD-L1 agents will gain approval in many additional malignancies, including in patients with locally advanced or newly diagnosed metastatic disease as was recently witnessed in melanoma, and a subset of non–small cell lung cancer patients.12
In an environment that is without fiscal restraints, patients were offered such therapies with risk falling to the payer (government, insurer, employer, and to a lesser extent the patient).
In the OCM world, such behavior adds risk to the participating practice. In addition, in the setting of higher patient co-payments, the patient now also bears more of the risk. Bearing risk in community oncology practices translates into risks to personal income and the practice’s vitality, challenging the physician to weigh personal interests against patient interests. While capitated contracts and risk-bearing contracts have existed in the past, particularly in healthcare systems and primary care practices, the scale of the OCM and the prices associated with these agents provide very broad risk corridors not typically seen in primary care or pediatrics practice.
Challenges in Delivering I-O Therapy Rationally
It is a physician’s obligation to deliver therapy that will benefit the patient, and the current data supporting I-O therapy provides compelling evidence that it should be part of the therapeutic armamentarium in a variety of malignancies. Unfortunately, we are missing several important pieces of data that might help physicians deliver these therapies rationally. Table 2
lists questions that might help moderate the costs of I-O.
While there is an unprecedented number of I-O trials actively enrolling patients, essentially all these trials are looking for expanded use of PD-1, PD-L1, or CTLA-4–binding agents or novel I-O agents, either alone or with the approved agents. These trials, if positive, will only accelerate the rise of I-O use and costs in cancer care.
Ideally, a society should support an environment of innovation that improves the lives of its citizens, and simultaneously permits the astute use of its limited resources. Improving the value of the care we deliver to cancer patients is important. The shift to value-based care, concurrent with the innovations in immunology and the rapidly growing indications for I-O agents, will create significant challenges for the physicians and healthcare systems in the near term. Community-based oncologists will need to be particularly savvy in meeting the operational and financial demands of clinical transformation prescribed by value-based care, while maintaining the financial viability of their practices. Finally, manufacturers of I-O agents should share the responsibility of supporting value-based pricing. The market is becoming more crowded with duplicate I-O agents, and competitive pricing will lower the tension that these treatments inflict on the value equation.