The unsolved problem of antibiotics in immuno-oncology

The introduction of immuno-oncology in the past decade and a half revolutionized cancer care. The big idea is to harness the power of the immune system to kill cancer, instead of using cytotoxic chemotherapy agents. The most successful class of these agents, the PD-1 inhibitors (also known as checkpoint inhibitors), work by “releasing the brake” from immune cells. PD-1 inhibitors have brought many solid tumor cancer patients into complete remission, treating tumors where medicine previously had little to offer. In the process, they have become some of the best-selling drugs of all time.

In a surprise finding, patients who receive antibiotics at the beginning of their PD-1 course of treatment find that the PD-1 no longer works against cancer. The rationale: immuno-oncology drugs rely on a healthy immune system. Antibiotics damage the gut microbiome, and the tight linkage between microbiome and immune system means that a damaged gut microbiome leads to a dysfunctional immune system.

As the oncology world moves increasingly towards more sophisticated immunotherapies, and more patients receive these treatments, a non-antibiotic treatment for bacterial infection will become increasingly important. We do not want to find ourselves in a position where cancer patients receive highly effective, modern oncology therapies but succumb to infection treated with antibiotics developed in the middle of the 20th Century.

To be more concrete with the evidence, we start with the clinical observation that patients receiving antibiotics up to 2 months prior and as late as 1 month after starting their immuno-oncology regimen show remarkably different survival outcomes. For the sake of illustration, one of the more dramatic study results is shown in Figure 14, taken from the work of Pinato and colleagues.¹⁷² These findings have been consistently repeated across the solid tumor cancer types treated with PD-1 inhibitors. A meta-analysis that incorporated 41,663 patients found a consistent, pronounced immuno-oncology impairment in both overall survival and progression-free survival in non-small cell lung cancer, urothelial cancer, melanoma, renal cell carcinoma, hepatocellular carcinoma, and others.¹⁷³ Results like this underline a strong and consistent clinical observation of immuno-oncology impairment correlated with antibiotic therapy at the beginning or two months before checkpoint inhibitor therapy.

The correlation is striking, but an obvious possible confounding variable is that patients who receive antibiotics are often sicker, with more comorbidities, than those who do not. The difference in survival could simply be that the antibiotic subgroup of patients is more frail. To evaluate correlation versus causation in this setting, Genentech studied the results from their Phase 3 programs for atezolizumab, their checkpoint inhibitor. In these studies, they had compared efficacy of atezolizumab head-to-head with the chemotherapy drug docetaxel in non-small cell lung cancer patients. They reasoned that patients in both groups would have received antibiotics. If the antibiotic subgroup is simply a sicker group of patients, then these patients should succumb faster in the chemotherapy group, too. What they found was the opposite: a non-significant difference in the chemotherapy group, but a significant difference in the atezolizumab group (see Table 2).¹⁷⁴ This finding supports the hypothesis that antibiotics interfere mechanistically with checkpoint inhibitors. It has been replicated in other major studies,¹⁷⁵ including a prospectively evaluated study in gastric cancer patients (see Figure 15).¹⁷⁶

Furthermore, laboratory studies in mice recapitulate the effect and build the case for a mechanistic understanding of the link between a healthy microbiome and a functioning immune system. In a study conducted at a leading PD-1 lab at Dana Farber, mice required a healthy microbiome for their checkpoint inhibitor therapy to control tumor growth (see Figure 16).¹⁷⁷

Tightening the causal chain, the microbiomes of PD-1 responders carry distinct signatures from non-responders, relatively depleted in some species and enriched in others (see Figure 17).^20,21,178 The non-responders lack strains corresponding to those most susceptible to antibiotics, primarily anaerobes.²¹ Furthermore, not all antibiotics have a negative impact on PD-1 outcomes, but rather those with broader spectrum and those more likely to deplete anaerobes.²¹ These differences carry through to peripheral T cell populations, linking antibiotic-induced anaerobe depletion to reduced T cell function (see Figure 18).²¹ Mechanistically, this helps explain why antibiotics can abrogate immuno-oncology therapeutic agents reliant on functioning T cells. The detailed workings of the mechanism underlying the gut-immune linkage remains an active field of investigation as researchers seek to understand the molecular basis of the gut-immune relationship.

About 150,000 US patients every year receive checkpoint inhibitors for their cancer diagnoses.^*179 Their PD-1 treatment typically costs $150,000 per patient.¹⁸⁰ About 30% of these patients require antibiotic treatment, which measurably worsens their cancer.^20,173 From a financial standpoint, treating these patients with generic antibiotics costs a few dollars and wastes tens or hundreds of thousands already spent on their cancer care.

Solid tumor patients on PD-1 therapy have well-documented, elevated rates of community infection driven by their underlying disease and demographics. Non-small cell lung cancer patients are at disproportionate risk for respiratory infection. Post-menopausal women (the majority of female PD-1 recipients) face chronically elevated UTI risk.^181–183 See Table 3 for common infections contracted by PD-1 patients.

Infectious disease guidelines explicitly classify malignancy as a comorbidity that escalates empiric antibiotic selection.^30,184,185 A community infection that would receive narrow-spectrum treatment in a healthy adult receives broader, more potent coverage in a cancer patient, by guideline.

We envision that bacterial defense disablers could be the modern infection control medicine that complements immuno-oncology. Like the PD-1 medicines, bacterial defense disablers rely on the exquisite power of the immune system to identify pathogens. Like the PD-1 inhibitors with cancer, bacterial defense disablers make the immune system more able to clear infection. Because bacterial defense disablers do not damage the microbiome, they would not interfere with the gut-immune axis necessary for PD-1 efficacy. Bacterial defense disablers could be given in response to an infection, or more likely, could be given in a preventive sense, in the way that colony stimulating factors are already given to chemotherapy patients to reduce infection risk. Colony stimulating factors are a cornerstone of many chemotherapy treatment protocols, allowing patients to be treated in outpatient settings because of their reduced infection risk. As such, they collectively provide more than $4B in revenue each year. Immuno-oncology needs a new paradigm for infection control. The product that meets that need will be valuable for patients and shareholders alike.