Inhibition of the CD47-SIRP alpha axis for cancer therapy: A systematic review and meta-analysis of emerging clinical data

CD47-SIRP alpha interaction acts as a "don't eat me " signal and is exploited by cancer to downregulate innate and adaptive immune surveillance. There has been intense interest to develop a mechanism of blockade, and we aimed to analyze the emerging data from early clinical trials. We performed a systematic review and meta-analysis of relevant databases and conference abstracts including clinical trials using CD47 and/or SIRP alpha inhibitors in cancer treatment. Nonlinear mixed models were applied for comparison of response and toxicity. We retrieved 317 articles, 24 of which were eligible. These included 771 response-evaluable patients with hematologic (47.1%) and solid tumors (52.9%). Of these, 6.4% experienced complete response, 10.4% partial response, and 26.1% stable disease for a 16.7% objective response rate (ORR), 42.8% disease control rate, and 4.8-month median duration of response. ORR was significantly higher for hematologic cancers (25.3%) than solid cancers (9.1%, p=0.042). Comparing by mechanism, seven CD47 monoclonal antibodies (mAbs) and six selective SIRP alpha blockers were given alone or combined with checkpoint inhibitors, targeted therapy, and/or chemotherapy. In solid cancers, selective SIRP alpha blockade showed a higher ORR (16.2%) than anti-CD47 mAbs (2.8%, p=0.079), which was significant for combination therapies (ORR 28.3% vs 3.0%, respectively, p=0.010). Responses were seen in head and neck, colorectal, endometrial, ovarian, hepatocellular, non-small cell lung, and HER2+gastroesophageal cancers. Dose-limiting toxicity (DLT) was seen in 3.3% of patients (5.4% anti-CD47 mAbs, 1.4% selective SIRP alpha blockers; p=0.01). The frequency of treatment-related adverse events (TRAEs) & GE;grade 3 was 18.0%, similar between the two groups (p=0.082), and mostly laboratory abnormalities. For anti-CD47 mAbs, the most common toxicities included grade 1-2 fatigue (27.2%), headache (21.0%), and anemia (20.5%). For selective SIRP alpha blockers, these included grade 1-2 infusion reaction (23.1%) and fatigue (15.8%). Anti-CD47 mAbs were significantly more likely than selective SIRP alpha blockers to cause grade 1-2 fever, chills, nausea/vomiting, headache, and anemia. In conclusion, combination therapies using selective SIRP alpha blockade had higher response rates in solid tumors than anti-CD47 mAb combinations. Hematologic changes were the main TRAEs, and selective SIRP alpha blockers seemed to have a better grade 1-2 toxicity profile. Treatment was well-tolerated with minimal DLTs.