Biomarker Hunt Focuses on Predicting Immunotherapy Adverse Events

OncologyLive, Vol. 19/No. 24, Volume 19, Issue 24

Although the need for biomarkers of immunotherapy response has generated much attention, investigators also are pursuing mechanisms for distinguishing which patients will experience adverse effects from these therapies.

Douglas B. Johnson, MD

Findings from recent studies have begun to answer a question that oncologists have been asking since the FDA approved the first immune checkpoint inhibitor, ipilimumab (Yervoy), in 2011: Which patients will tolerate a particular immunotherapy and which patients will suffer serious toxicities?

Investigators have identified several biomarkers that are significantly associated with immune-related adverse events (irAEs) in clinical trial data, both for patients with melanoma who use certain checkpoint inhibitors and in patients with leukemia who receive T cells modified with chimeric antigen receptors (CARs).

The biomarkers found to date are far from perfect prediction tools, even among patients in the studies in which they were analyzed. Additional studies are needed to identify more predictive biomarkers and to validate those found to date, experts say. As immunotherapy becomes more widely used and information from hundreds of ongoing trials accumulates, far more data will be available for analysis.

“There is a great deal of interest right now in finding out why only certain people experience [adverse] effects [AEs] and the related question of why only certain people respond to immunotherapies. The National Cancer Institute is investing considerable money in research, and a number of groups are investigating,” said Douglas B. Johnson, MD, clinical director of the melanoma research program at Vanderbilt University Medical Center in Nashville, Tennessee.

“Ideally, we’d like to find biomarkers that accurately predict which patients will experience serious [adverse] effects, and even if we find no predictive biomarkers, the work we’re doing should help us understand the pathways through which [adverse] effects develop, and that could give us better ways to prevent or treat those effects…But we’re a long way from that. We’re just beginning to get a little peek into what has been a black box.”

Search for Answers Starts With Autoimmunity

Grade 3 or higher irAEs are observed in up to 43% of all patients who take ipilimumab, the only approved CTLA-4 inhibitor. Drugs that target PD-1 and PD-L1 are better tolerated, producing grade 3 or higher irAEs in up to 20% of patients.1 Combination therapy results in more frequent irAEs than either type of checkpoint inhibitor on its own.2 Although predictive biomarkers for these effects have not been established, the Society for Immunotherapy of Cancer has recommended a list of tests for oncology specialists to consider before prescribing immune checkpoint inhibitor therapy (Table 1).1The most common irAEs associated with checkpoint inhibitors are autoimmune attacks on healthy cells, so researchers hypothesized that preexisting autoimmune disease would be among the best predictors of new irAEs; these patients have routinely been excluded from immunotherapy trials. However, investigators recently found that although an underlying autoimmune disorder can flare during ipilimumab therapy or can result in irAEs, a preexisting condition does not negatively affect response, and the resulting irAEs are manageable.3,4

Studies focused on less-intuitive baseline predictors initially did not prove much more fruitful. In a 2011 study of ipilimumab therapy, investigators made pairwise comparisons of serum interleukin-17 (IL-17) levels in patients with colitis (n = 13) versus those with no irAEs (n = 16). Serum IL-17 levels were significantly higher in patients with colitis at weeks 7 (P = .007) and 12 (P = .02). However, although IL-17 levels correlated with the development and resolution of colitis, pretreatment IL-17 levels were similar between patients with or without colitis, invalidating IL-17 as a credible predictive factor.5

A 2013 study performed gene expression profiling on 162 patients who took ipilimumab for advanced melanoma. Baseline samples showed 27 probe sets—most of them related to patient immune system, cell cycle, and intracellular trafficking—with differential mean expression (≥1.5-fold; P ≤.05) in patients who did and did not experience gastrointestinal irAEs. However, the investigators concluded that the low sensitivity of the biomarkers they found would prevent them from being used on their own to predict irAEs.6

Table 1. Pretreatment Evaluation and Diagnostic Tests to Consider in Patients Prior to Initiating Checkpoint Inhibitor Therapy1

Signals Emerge

A 2014 study retrospectively analyzed baseline, week 4, and week 7 relative and absolute eosinophil count (AEC), relative eosinophil count (REC), and subsequent irAEs in 156 patients who took ipilimumab for advanced melanoma. Changes in AEC and REC levels were both significantly associated with irAEs, but baseline readings alone were not.7Several more recent attempts to find baseline biomarkers have fared better. Investigators have found significant associations between several starting factors and the eventual onset of irAEs among some patient populations who use immune checkpoint inhibitors.

A 2015 study of ipilimumab in 35 patients with melanoma tested for 36 functionally selected cytokines and chemokines at baseline and week 6. Unlike the 2011 study, the investigators found that baseline levels of IL-17 were significantly associated with the incidence of grade 3 diarrhea or colitis (P = .02). There was no significant association between baseline IL-17 and all-grade irAEs.8

A 2017 study compared baseline body characteristics with ipilimumab irAEs and found that patients with sarcopenia and low muscle attenuation (MA), assessed using computed tomography scans, were more likely to experience severe events. The study included 84 patients, 24% of whom were sarcopenic and 33% of whom had low MA at baseline. Multivariate analysis found that sarcopenia and low MA were significantly associated with high-grade irAEs, with odds ratios (ORs) of 5.34 (95% CI, 1.15-24.88; P = .033) and 5.23 (95% CI, 1.41-19.30; P =.013), respectively. Additionally, low MA was associated with high-grade irAEs (OR, 3.57; 95% CI, 1.09-11.77; P = .036). Patients with sarcopenia were more susceptible to high-grade irAEs but the rate was not statistically significant.9

Gene Signature

In a 2018 study, investigators analyzed wholeblood samples from 150 patients with stage III/IV melanoma who received tremelimumab, an experimental CTLA-4 inhibitor, during 2 clinical trials to determine whether they could identify genes that would predict grade 3/4 diarrhea and colitis, the most frequent irAEs leading to treatment discontinuation. They looked at 169 genes associated with inflammation, immunity, and melanoma.

No single gene in baseline blood samples was associated with diarrhea or colitis. However, a 16-gene signature in blood taken 30 days after the start of treatment distinguished patients who developed grade 2 or higher diarrhea or colitis from those who would experience grade 0/1 severity of these irAEs (Table 2).10 The investigators validated the signature with samples from another 210 patients and found that it differentiated patients who developed grade 0/1 diarrhea or colitis from those who had grade 2 to 4, with an area under the curve of 0.785 (95% CI, 0.723-0.838, P <.0001), a sensitivity of 57.1%, a specificity of 84.4%, a positive predictive value of 57.1%, and a negative predictive value of 84.4%.10

“We do not validate a signature that identifies all patients who will develop diarrhea severe enough to require treatment discontinuation. However, our data suggests that this minimally invasive strategy can identify early in a given patient’s treatment course those at risk for the development of high-grade diarrhea or colitis,” Philip Friedlander, MD, PhD, and colleagues wrote in the Journal for Immunotherapy of Cancer. “This information potentially could allow for early intervention strategies that would limit immune-related adverse event severity or prevent toxicity development and thereby extend treatment duration.”10

Antibody Profiling

In another 2018 study, investigators from NYU Perlmutter Cancer Center in New York, New York, used a proteome array to analyze baseline antibody levels in patients with advanced melanoma to search for biomarkers associated with AEs for ipilimumab and the anti—PD-1 inhibitors pembrolizumab (Keytruda) and nivolumab (Opdivo).

Investigators collected pretreatment sera samples from 37 patients who received ipilimumab, 27 who took an anti—PD-1 inhibitor as monotherapy, and 11 who underwent combination therapy with ipilimumab and an anti–PD-1 medication. They found hundreds of antibodies associated with severe toxicity that were differentially expressed (DE) in patients with mild or no toxicities (P <.05). There were 914 DE antibodies in the anti—CTLA-4 cohort, 723 in the anti–PD-1 group, and 1161 in those who received combination immunotherapy. Investigators also determined that there was little overlap between unique DE antibodies and either anti–CTLA-4 or anti–PD-1 monotherapy and combination therapy with both of the immune checkpoint inhibition agents (Figure).11

The team then studied the potential role of toxicity-associated antibodies in irAEs by analyzing the protein antigen targets related to these antibodies in immune-related pathways such as interleukin-1 (IL-1), toll-like receptor signaling, Escherichia coli infection, and micro-RNA biogenesis. Focusing on the 15 most DE toxicity—associated antibodies, they found that their corresponding protein targets were highly expressed in liver and skin tissue and identified with immune cell activity and autoimmune disorders.11

The findings showed that patients with metastatic melanoma with a specific autoantibody profile are more likely to develop severe irAEs, the investigators concluded. They added that the results would have to be validated prospectively before serum antibody biomarkers could be used as a predictive tool. At the same time, the researchers noted the importance of developing such information so that patients receiving immune checkpoint therapy could be monitored more closely and, if necessary, have their treatment regimen modified or receive earlier prophylactic agents, such as corticosteroids.

Table 2. Gene Signatures for Predicting Immune-Related Adverse Events10

Translating the Findings

“Ultimately, validation of these predictive biomarkers could enable clinicians to optimize the risk—benefit assessment for individual patients to maximize therapeutic benefit while minimizing possible severe toxicities from immune checkpoint inhibitors,” Gowen et al wrote.Although the discovery and validation of toxicity-predicting biomarkers are viewed as important research goals, experts have different views on how such knowledge would affect clinical practice.

“The problem with knowing in advance about checkpoint inhibitor [adverse] effects is that—except in the case of effects that cannot be managed effectively and lead to severe disability or death, which occurs in just 0.4% of patients with PD-1 inhibitors and 1.2% for the ipilimumab-nivolumab combination—the knowledge would not change our management. That is, we still would use the checkpoint inhibitor, even with the knowledge that a patient is likely to develop pneumonitis or a similar [adverse] effect,” said Igor Puzanov, MD, MSCI, FACP, director of the Early Phase Clinical Trials Program at Roswell Park Comprehensive Cancer Center in Buffalo, New York.

The reasoning is that AEs that are caught early can usually be managed effectively—generally with topical steroids for skin toxicity and corticosteroids for internal toxicity12—but the benefits that immunotherapies provide are hard to replace. Indeed, to take advantage of a tool that predicts AEs, “one has to have an alternative, similarly effective approach to therapy,” Puzanov added. “Such alternatives may exist in some situations—BRAF/MEK targeting versus checkpoint inhibitor in a patient with BRAF-mutant melanoma—but not always. And the substitution might not even be justified in BRAF-mutant melanoma, except for the subgroup of patients with good performance score, low disease burden, and low LDH [lactate dehydrogenase]. In many other situations, such a choice is nonexistent, BRAF wild-type patients with melanoma being such an example. And even the alternatives may have [adverse] effects.”

In theory, predictive biomarkers might allow doctors to monitor some patients more closely than others, but Puzanov says that it would be difficult to use biomarkers to provide extra surveillance, because current guidelines recommend very close surveillance for all patients, and then use biomarkers to provide less surveillance to other patients, unless the biomarkers predicted all AEs. “Unless the test has a 100% positive or negative predictive value, one would have to continue to monitor patients in a very similar fashion,” he said.

Figure. Antibody Profiling in Immune Checkpoint Therapy11

CAR T-Cell Therapy Biomarkers

The value of biomarkers could increase, however, as immune checkpoint inhibitors gain approvals as first-line treatments for patients who do have other effective options. “Checkpoint inhibitors are already in trials as frontline treatment for lung cancer and kidney cancer, and there are other active frontline treatments for both of those conditions,” said Johnson, the Vanderbilt expert. “If we discovered biomarkers that predicted which lung or kidney cancer patients would have severe immune-related [adverse] effects, those biomarkers could justify starting some patients with other treatment types.”Efforts to find biomarkers that predict AEs from the second major type of immunotherapy, CAR T-cell therapy, are at a similar point. Investigators have found some interesting candidates, but they have yet to validate their findings or apply them to patient care.

CAR T-cell therapy is associated with 2 major types of AEs: cytokine release syndrome (CRS) and neurological toxicities. Most patients who receive CAR T-cell therapy experience some degree of CRS, which occurs when T cells release cytokines that cause a variety of symptoms, including fever, nausea, headache, rash, rapid heartbeat, low blood pressure, and breathing difficulty. The condition becomes serious to the point of life-threatening when the first set of cytokines stimulates the release of more cytokines, which then spur the release of more cytokines and so on. Trials of CD19-specific CAR T-cell therapies for relapsed/refractory B-cell acute lymphoblastic leukemia (ALL) have reported that incidence rates of grade 3, 4, or 5 CRS ranged from 19% to 43% of patients.12

Unlike CRS, which typically occurs within the first few weeks after patients begin treatment, neurological toxicities typically occur later. All patients are at risk, but patients who first develop CRS are more likely than other patients to develop neurological toxicities, which include headaches, confusion, alterations in wakefulness, hallucinations, dysphasia, ataxia, apraxia, facial nerve palsy, tremor, dysmetria, and seizures. The size of the risk is hard to quantify, however. CAR T-cell trials have reported incident rates of neurological toxicity that have ranged from 0% to 50%.13

Results from a 2017 study of 19-28z CAR T cells in 51 patients with CD19-positive B-cell ALL found several clinical and serum biomarkers that were associated with the subsequent development of severe neurotoxicity. Investigators performed multivariate analysis on the 21 patients who experienced severe neurological toxicity and the 30 who did not, and they found that such AEs were significantly associated with high disease burden. That was defined as ≥50% blasts at the time of T-cell infusion (P = .0045) and with posttreatment CRS grade ≥3 (P = .0010). Severe neurotoxicity was also associated with peak serum cytokines at day 3.14 Other studies have found that neurotoxicity is more common in younger patients, nonheavily pretreated patients, and, possibly, those with preexisting neurological conditions.15

Additionally, study findings have shown that high baseline disease burden is a significant risk factor for CRS as well as for neurotoxicity. Indeed, recent research results demonstrate that preexisting endothelial activation and high-intensity lymphodepletion with fludarabine are associated with higher risk for both AEs. Higher CAR T-cell doses may also increase risk, although study findings are not conclusive.16

“We are getting significantly better at managing adverse events that stem from CAR T cells, and the cell therapy community is doing some things in trials now that we hope will make us better still—such as administering tocilizumab [Actemra] at the first sign of fever—that we hope will prevent CRS from progressing to severe in some cases,” said Eric Smith, MD, PhD, director of clinical translation in the Cellular Therapeutics Center at Memorial Sloan Kettering Cancer Center in New York, New York.

“Still, having a good idea of which patients face the highest risk of severe [adverse] effects could certainly be beneficial. For example, we have experience on a clinical trial giving lower doses to patients with the highest tumor burden in hopes of maintaining therapeutic benefit while avoiding the overstimulation that may create adverse events in those patients…Knowing who faces the greatest risk of [adverse] effects would also help efforts to move CAR T-cell therapy from a procedure that requires hospitalization to one that can be delivered on an outpatient basis, at least for low-risk patients.”

References

  1. Puzanov I, Diab A, Abdullah K, et al; Society for Immunotherapy of Cancer Toxicity Management Working Group. Managing toxicities associated with immune checkpoint inhibitors: consensus recommendations from the Society for Immunotherapy of Cancer (SITC) Toxicity Management Working Group. J Immunother Cancer. 2017;5(95). doi: 10.1186/s40425-017-0300-z.
  2. Winer A, Bodor JN, Borghaei H. Identifyinng and managing the adverse effects of immune checkpoint blockade. J Thorac Dis. 2018;10(suppl 3):S480-S489. doi: 10.21037/jtd.2018.01.111.
  3. Gutzmer R, Koop A, Meier F, et al; German Dermatooncology Group (DeCOG). Programmed cell death protein-1 (PD-1) inhibitor therapy in patients with advanced melanoma and preexisting autoimmunity or ipilimumab-triggered autoimmunity. Eur J Cancer. 2017;75:24-32. doi: 10.1016/j.ejca.2016.12.038.
  4. Johnson DB, Sullivan RJ, Ott PA, et al. Ipilimumab therapy in patients with advanced melanoma and preexisting autoimmune disorders. JAMA Oncol. 2016;2(2)234-240. doi: 10.1001/jamaoncol.2015.4368.
  5. Callahann MK, Yang A, Tandon s, et al. Evaluation of serum IL-17 levels during ipilimumab therapy: correlation with colitis. J Clin Oncol. 2011;29(suppl 15):2505. ascopubs.org/doi/abs/10.1200/jco.2011.29.15_suppl.2505.
  6. Shahabi V, Berman D, Chasalow SD, et al. Gene expression profiling of whole blood in ipilimumab-treated patients for identification of potential biomarkers of immune-related gastrointestinal adverse events. J Transl Med. 2013;11:75. doi: 10.1186/1479-5876-11-75.
  7. Schindler K, Harmankaya K, Kuk D, et al. Correlation of absolute and relative eosinophil counts with immune-related adverse events in melanoma patients treated with ipilimumab. J Clin Oncol. 2014;32(suppl 15):9096. ascopubs.org/doi/abs/10.1200/jco.2014.32.15_suppl.9096.
  8. Tarhini AA, Zahoor H, Lin Y, et al. Baseline circulating IL-17 predicts toxicity while TGF-β1 and IL-10 are prognostic of relapse in ipilimumab neoadjuvant therapy of melanoma. J Immunother Cancer. 2015;3:39. doi: 10.1186/s40425-015-0081-1.
  9. Daly LE, Power DG, O’Reilly Á, et al. The impact of body composition parameters on ipilimumab toxicity and survival in patients with metastatic melanoma. Br J Cancer. 2017;116(3):310-317. doi: 10.1038/bjc.2016.431.
  10. Friedlander P, Wood K, Wassmann K, Christenfeld AM, Bhardwaj N, Oh WK. A whole-blood RNA transcript-based gene signature is associated with the development of CTLA-4 blockade-related diarrhea in patients with advanced melanoma treated with the checkpoint inhibitor tremelimumab. J Immunother Cancer. 2018;6(1):90. doi: 10.1186/s40425-018-0408-9.
  11. Gowen MF, Giles KM, Simpson D, et al. Baseline antibody profiles predict toxicity in melanoma patients treated with immune checkpoint inhibitors. J Transl Med. 2018;16(1):82. doi: 10.1186/s12967-018-1452-4.
  12. Bonifant CL, Jackson HJ, Brentjens, RJ, Curran KJ. Toxicity and management in CAR T-cell therapy. Mol Ther Oncolytics. 2016;3:16011. doi: 10.1038/mto.2016.11.
  13. Brudno JN, Kochenderfer JN. Toxicities of chimeric antigen receptor T cells: recognition and management. Blood. 2016;127(6):3321-3330. doi: 10.1182/blood-2016-04-703751.
  14. Park J, Riviere I, Wang X, et al. Baseline and early post-treatment clinical and laboratory factors associated with severe neurotoxicity following 19-28z CAR T cells in adult patients with relapsed B-ALL. J Clin Oncol. 2017;35(suppl 15):7024. ascopubs.org/doi/abs/10.1200/JCO.2017.35.15_suppl.7024.
  15. Wang Z, Han W. Biomarkers of cytokine release syndrome and neurotoxicity related to CAR-T cell therapy. Biomark Res. 2018;6:4. doi: 10.1186/s40364-018-0116-0.