Xiaomo: Kangfang Biology (09926) HarmonI-2 overall survival data strongly maintains “gain” rating

The Zhitong Finance App learned that J.P. Morgan Chase released a research report saying that Kang Fang Biotech (09926) announced an update to overall survival (OS) data. The overall OS data is quite strong. The risk ratio (HR) is 0.73, the confidence range is 0.57 to 0.95, and the p-value is 0.009. The median OS in the ivonescimab (ivonescimab) group was 30.8 months, which was more than 8 months longer than 22.6 months compared to the pembrolizumab (pembrolizumab) group. The bank maintains Kangfang Biotech's “gain” rating. The target price is HK$125. Based on the December 2027 valuation, the cash flow discount method is used, assuming a terminal growth rate of 3% and a weighted average capital cost of 9.6%.

The bank pointed out that the median OS of pabolizumab in Harmoni-2 was about two months longer than the median OS of pabolizumab in the Keynote-042 Chinese study. Considering that many new NSCLC treatments, including ADC, have appeared since KEYNOTE-042 began in 2014, this result is not surprising. The safety profile of the two groups was roughly similar. For example, the ratio of serious treatment-related adverse events (TRAE) in the evosimab/pabolizumab group was 29.9%/21.6%, respectively, and the TRAE ratio leading to permanent discontinuation was 4.1%/5%, respectively.

The bank believes that the Harmoni-2 OS data update will have positive implications for two global phase III trials, namely the Harmoni-3 squamous cell cohort and Harmoni-7, respectively. In HarmonI-2, the OS HR for squamous cell patients was 0.65, far lower than the OS HR of 0.79 for non-squamous cell patients; furthermore, the OS HR for PD-L1 TPS = 50% patients was 0.58, which is the target patient group for Harmoni-7. The bank indicated that the strong OS HR of these two subgroups made them more confident in reading the Harmoni-3 squamous cell cohort and Harmoni-7 OS.

The research report also clarified several aspects of the HarmonI-2 statistical plan, including the current WCLC data as a result of the mid-term analysis of the second plan. The previous OS data (OS HR=0.777) was the result of the first mid-term analysis, and alpha consumption was minimal. The default number of events for the second mid-term analysis was 232, and the one-sided alpha was 0.0139. As of the data cutoff date of August 20, 2026, the median follow-up time was about 36 months, 234 OS incidents occurred, and the one-sided alpha was 0.0141. At the time of the first mid-term OS analysis, it was assumed that in February 2025, the median follow-up time was only about 17 months, far lower than the median OS of the two groups, so the OS HR 0.777 for the first mid-term analysis was quite immature.