In the teaching lecture on ‘Immunotherapy in Melanoma: From Frustration to Enthusiasm’ at ECCO-ESMO’s 2015 annual conference, Professor Jürgen C Becker referred to the latest immunotherapy developments as “It’s not just a wave, it’s a tsunami.” Professor Becker reminded the audience that costly immunotherapies, now approved for the few thousand patients with melanoma, will also likely be available for cancer indications with a larger incidence, putting a huge financial strain on society and underscoring the question: How can we ensure immunotherapies are used appropriately? Indeed, the many recent exciting developments for immunotherapy, specifically for immune checkpoint inhibitors, in cancer have increased enthusiasm for these agents but have also brought to light many more questions that now more than ever need to be addressed. As Dr. James Larkin, a Consultant Medical Oncologist at The Royal Marsden, commented in the ‘Immunotherapy in Melanoma’ Satellite Symposium, “We need time for the dust to settle.”
Employing the body’s immune system to fight cancer is not a new approach. As outlined by Professor Caroline Roberts (head of the Dermatology Unit at the Institut Gustave-Roussy, Paris) in the opening plenary session, Bacillus Calmette-Guérin (BCG) has been used for the treatment of bladder cancer for more than 30 years, IL-2 was approved for the treatment of renal cell carcinoma (RCC) in the early 1990’s, numerous vaccine trials were carried out in the 1990s, and adoptive T-cell transfer has been investigated in many trials from the year 2000. However, the real breakthrough in the immunotherapy field came with discovery that blockade of a single immune checkpoint inhibitor can elicit profound clinical effects.
Ipilimumab (Bristol Myer-Squibb’s Yervoy) was the first immune checkpoint inhibitor (targeting CTLA-4) to gain regulatory approval for cancer (in 2011 for melanoma), but the discovery that blockade of PD-1 signalling could exert even greater clinical benefit has thrown immunotherapy, with some force, back into the spotlight, with this years’ ECCO-ESMO annual conference devoting extensive coverage on the latest developments of immune checkpoint inhibitors in oncology. Pembrolizumab (Merck & Co.’s Keytruda) and nivolumab (Bristol Myer-Squib’s Opdivo) are the only two FDA- and EMA-approved anti-PD-1 therapies (for melanoma). However, they are being explored in multiple tumor types; pembrolizumab has demonstrated activity in bladder cancer, head and neck cancer, urothelial cancer, gastric cancer, breast cancer, non-Hodgkin’s lymphoma, and lung cancer among others. But as the number of indications in which immune checkpoint inhibitors are being evaluated expands and data from more clinical trials is reported, the degree of perplexity surrounding immune checkpoint inhibitor seems to be increasing.
Presentation of the late-breaking abstract (abstract 3LBA) for the Phase III Checkmate-025 trial of nivolumab versus everolimus in pre-treated metastatic renal cell carcinoma (RCC) was eagerly anticipated and the data, presented in the Presidential Session on September 26, did not disappoint. Professor Padmanee Sharma, of the MD Anderson Cancer Centre, Texas, presented data illustrating the first significant OS benefit in a Phase III trial metastatic RCC—25.0 months for nivolumab vs. 19.6 months for everolimus. The trial met its primary end point of median OS and was stopped early to allow patient crossover. Although there was no significant difference in PFS at the time of follow-up, the PFS curves separated late indicating potential for a delayed PFS benefit. Most importantly, the survival curves separated early and remained separated. Of note, the OS benefit was observed regardless of PD-L1 expression. In the discussion following presentation of Checkmate-025, led by Professor Sternberg, several questions critical to successfully integrating nivolumab into the clinical practice for RCC were raised, specifically when to stop therapy given that in some patients a best response of stable disease can lead to improved survival and that other patients experience a delay in tumor response to treatment.
The question of how long to treat patients with an immune checkpoint inhibitor was also brought to our attention in the presentation of a Phase II trial of nivolumab in metastatic RCC (abstract 501), in which patients were treated with nivolumab beyond disease progression. The rationale for treating beyond progression is that ‘tumor flare’ may be seen with immunotherapy treatment, where infiltrating T-cells to the tumor site result in apparent disease progression by imaging as defined by RECIST. In this trial, treatment beyond disease progression was permitted if nivolumab was tolerated and clinical benefit noted, which is in itself open to interpretation. The data showed that in some patients who continued nivolumab treatment, subsequent extended tumor shrinkage and extended survival was observed (for more information please see the companion ESMO blog from I. Ammermann entitled, “Batman (Opdivo) and Robin (Cometriq) Come to Rescue Renal Cell Carcinoma”.
In the ‘Immunotherapy in Cancer’ proffered paper session on September 27, the conference hall was full to capacity. Data from three late-breaking abstracts, for antibodies this time targeting the ligand of PD-1, PD-L1, in non-small cell lung cancer (NSCLC), were presented:
- The first from the Phase III POPLAR study of previously treated patients with PD-L1-positive squamous and non-squamous NSCLC (abstract 14LBA).
- The second for correlation of response to durvalumab with PD-L1 expression and tumoral IFN-γ mRNA (abstract 15LBA).
- The third from the single-arm Phase II BIRCH study of atezolizumab in PD-L1-positive NSCLC (untreated and previously treated patients) (abstract 16LBA). It is noteworthy to mention that atezolizumab in the subject of five ongoing trials in NSCLC.
In the POPLAR trial, atezolizumab improved OS compared with docetaxel (12.6 months compared with 9.7 months, at a minimum of 13 months follow-up the indications are that curves will remain separated. Patients were stratified by PD-L1 expression on tumor cells and immune cells as assessed by the Ventana SP142 anti-PD-L1 antibody in an immunohistochemistry (IHC) assay, separating expression into four levels. Unlike the data presented for RCC, an increased benefit in OS and PFS was observed in NSCLC with increased PD-L1 expression (both tumor cell expression and immune cell expression of PD-L1 were independent predictors of response). In addition, an increased benefit was observed in non-squamous NSCLC compared with squamous. The SP142 IHC assay was also used to assess tumor cell and immune cell expression of PD-L1 in the BIRCH study. The 6-month OS rate for atezolizumab was consistent with that observed in the POPLAR study and once again there was greater benefit (in terms of OS and ORR) in patients with higher levels of PD-L1 expression. A different IHC assay was used to assess PD-L1 expression in the durvalumab study (using the Ventana SP263 antibody), furthermore only tumor cells were assessed and there was a single cut-off for PD-L1-positivity (≥ 25% intensity). In addition to PD-L1 as a predictor for response to durvalumab, levels of tumoral IFN-γ mRNA were also found to positively correlate with tumor response, with greatest response observed in double PD-L1- and IFN-γ-positive patients.
These presentations raised yet more questions surrounding the use of immune checkpoint inhibitors. What assay should be used to assess PD-L1 expression? Should tumor cells, immune infiltrating cells, or both be tested to determine PD-L1-positivity? What level of expression of PD-L1 is needed for a patient to be considered PD-L1-positive? These data also raise the possibility of combining another biomarker with PD-L1 expression to improve its predictive value.
Another question raised during the cancer conference was, why do some cancers but not others respond to anti-PD-1 therapy? Data from the mulitcohort Phase Ib KEYNOTE-028 trial of pembrolizumab in PD-L1-positive advanced solid tumors were presented; patients with heavily pretreated squamous cell carcinoma of the anal canal responded to pembrolizumab (abstract 500) but not patents with microsatellite-stable colorectal cancer (abstract 502). PD-L1 expression was assessed by IHC using Merck’s 22C3 antibody clones and ≥ 1% PD-L1 expression in tumor cells and stroma was considered PD-L1-positive; another method of assessing PD-L1 expression to add to the mix.
Data from the Phase II IMvigor-210 trial of atezolizumab in locally advanced or metastatic bladder cancer (abstract 21LBA) were also presented. The number of attendees for this presentation exceeded expectation and the overflow area was required to accommodate the audience. Once again, higher PD-L1 status (SP142 IHC staining of immune cells only) was associated with greater response—an ORR of 27% in the IC2/3 subgroup compared with 18% in the IC1/2/3 PD-L1 subgroup and 9% in IC0 subgroup. Importantly, the responses were durable. Although positive, the trial also raises the question of whether tumor cells should also have been tested for PD-L1 expression (for more information please see the companion ESMO blog from J. Bamford entitled, “Back to the Future: 25 years after the approval of the BCG vaccine, immunotherapy is once again set to breathe new life into bladder cancer treatment” ).
Following a familiar theme for immunotherapy sessions, the first proffered paper session for melanoma on September 27 was packed to bursting point. Three presentations for immune checkpoint inhibitors were made:
- The first for pembrolizumab in advanced merkel cell carcinoma (MCC, a rare type of skin cancer) (abstract 22LBA).
- The second for the Phase III Checkmate-064 trial of nivolumab (given for six cycles) followed by ipilimumab (given for four cycles) and vice versa in advanced melanoma (abstract 23LBA).
- The third for the Phase Ib MASTERKEY trial of Amgen’s oncolytic virus talimogene laherparepvec (TVec) in combination pembrolizumab in unresectable stage IIIBIV melanoma (abstract 24LBA).
In MCC, an indication with no FDA-approved therapy, pembrolizumab (given for up to two years) induced tumor response in 71% of patients (10 out of 14) with profound tumor shrinkage, indicating yet another cancer patient population where immune checkpoint inhibitors appear to be beneficial. Comparable to other indications, the questions of which patients are most likely to response to treatment remain and further biomarker studies are ongoing to address the relationship of virus-positivity with response.
In Checkmate-064, there were consistent improvements in efficacy outcomes in patients treated first with nivolumab followed ipilimumab compared with patients receiving ipilimumab followed by nivolumab; at week 25, the ORR from the two treatment schedules was 47.7% and 22.6%, respectively. Interestingly, the rate of grade 3/4 adverse events after the two treatment periods was higher for the nivolumab followed by ipilimumab treatment arms (50% compared with 43%). These data raise the important question of whether patients should be treated with immune checkpoint inhibitors for a fixed duration rather than until disease progression. With no head-to-head trials investigating this in melanoma, cross trial comparisons of survival data will likely be used to inform this decision.
The data indicate nivolumab should be given as a first-line therapy ahead of ipilimumab, but what is the biology behind this finding? Professor Caroline Robert indicated she was surprised by these results—ipilimumab is involved in the primary phase of T-cell activation and recruitment whereas nivolumab is involved in the effector phase, therefore one might expect nivolumab to greater effect after ipilimumab treatment. Professor Robert also commented that there is no clear benefit of this sequencing compared with nivolumab therapy alone and that perhaps different doses of nivolumab and ipilimumab given sequentially in this way may have produced better efficacy results and a better toxicity profile.
Preliminary data from the MASTERKEY trial indicate activity of TVec in combination with pembrolizumab in melanoma and gives an example of immune checkpoint inhibitors being combined with another immunotherapy. However, the results seen with pembrolizumab monotherapy in melanoma (the comparator in the Phase III portion of the MASTERKEY trial) have raised the bar for TVec, Prof Dirk Schadendorf commented that the TVec combination has a “high hurdle to overcome”—patient selection will be critical for the Phase III portion of the trial to be success. Furthermore, this trial design raises the question of how will healthcare systems support the cost of such immunotherapy combination treatments?
Data for the combination of immune checkpoint inhibitors nivolumab and ipilimumab in the first-line treatment of metastatic melanoma (the Phase III Checkmate-067 trial) were presented by Dr. James Larkin in the second proffered paper session in melanoma (abstract 3303). Although presented previously, it was reiterated that 55% of patients treated with the combination experienced grade 3/4 toxicity compared with 16% of patients treated with for nivolumab alone and 27% for patients treated with ipilimumab alone. PFS analysis by subgroup indicated an overall trend for better outcome in the combination treated patients compared with nivolumab alone. The same trend was observed for ORR across all subsets. Interestingly, a high proportion of patients continued to have a response to nivolumab plus ipilimumab or nivolumab alone after treatment discontinuation (again raising the question of how long to treat patients). Despite the efficacy benefits of treating with the combination of nivolumab with ipilimumab, in a substantial proportion of these patients two or three organ systems were impacted by combination treatment (e.g. patients experienced gastrointestinal and hepatic adverse events at the same time), and some of these adverse events were experienced some time after treatment discontinuation making management of the side effects difficult.
Without a biomarker to select patients for treatment with nivolumab plus ipilimumab over nivolumab alone, physicians will face the difficult decision of whether to expose their melanoma patients to a toxic regimen that may or may not offer benefit. Survival data from Checkmate-67 will be critical in aiding this decision process.
Currently, we do not have an answer of how to select patients for treatment with immune checkpoint inhibitor therapy. PD-L1 continues to be investigated as a biomarker for anti-PD-1 therapies but there is an urgent need for a standardized PD-L1 assay and several questions need to be answered to implement PD-L1 testing in the clinic, such as:
- How should we define PD-L1-positive tumors?
- Should tumor cells and/or immune cells or analysed for PD-L1 expression?
- What assay and anti-PD-L1 antibody should be used?
- PD-L1 can be expressed constitutively and induced, so when is the best time to test?
- What tumor site should be tested; the primary tumor site, metastatic sites, or both?
Perhaps the most convincing reason that PD-L1 is not a suitable biomarker is that some patients, who are PD-L1-negative irrespective of the assay used, still respond to anti-PD-1 therapy. Indeed, PD-L1 expression completely loses its predictive value in melanoma patients treated with the combination of nivolumab with ipilimumab—patients who are PD-L1-negative seem to respond better to single agent nivolumab than to the combination. But are there any other potential biomarkers that could be used to select patients for anti-PD-1 therapy? It has been suggested that tumor mutational load and neoantigen load may provide predictive markers for immune checkpoint inhibitors, however this needs further investigation.
In the absence of biomarkers for anti-PD-1 therapy in melanoma, Dr. Larkin explained that we need to be smarter in how patients are selected for therapies, taking into consideration multiple factors including patient history, performance status, disease sites, LDH level, tumor BRAF status, and patients’ wishes. For example, a patient with a history of autoimmune disease should not be considered for immunotherapy. However, these set of criteria also need to be further defined.
This years’ ECCO-ESMO’s conference provided an information overload on the latest developments immune checkpoint inhibitors. However, there are many questions, some of which have been outlined here, that still need to be addressed before these therapies can successfully and importantly most appropriately be integrated into the clinical setting. The immune checkpoint inhibitor can of worms remains open.