Abstract
This letter to the editor responds to the letter from Su et al, regarding concerns related to immortal time bias that may partially account for recently published study results.
We very much appreciate the letter from Su and colleagues to The Oncologist.1 In this case, their concern is that the patients in the selected “AC group” (surgery first followed by postoperative chemotherapy) would have to survive longer than the patients in the “NAC group” (chemotherapy first followed by surgery) to be included in the analysis. Those who died before getting to surgery may not get counted, creating unfair survival advantages for the NAC group. Since patients treated with NAC had survival advantages over those treated with AC only in non-academic settings, we addressed this in 2 different settings: non-academic vs academic institutions.
We found that only 2 and 10 patients in academic setting and 1 and 10 in non-academic setting died within 90 and 180 days of chemotherapy, respectively. Therefore, those are very small numbers. We then performed 2 intention-to-treat (ITT) analyses in both settings. In the first ITT analysis (a), NAC patients received chemotherapy as the first course; they either had surgery or were recommended for surgery later. AC patients received surgery as the first course; they either had chemotherapy or were recommended for chemotherapy. Only 6 additional cases (4 for NAC and 2 for AC) were added compared to the original definition and the results were similar to the ones from the original definition. We also used broader criteria for a second ITT analysis (b). In analysis b, NAC patients receive chemotherapy but may or may not have received surgery as the first course treatment, whereas AC patients received surgery and may or may not have received chemotherapy as the first course treatment. One hundred fifty three additional AC cases and 140 additional NAC cases were added. The average lead-time from the date of diagnosis to the first treatment received in this analysis is AC:NAC 16.37 ± 27.90 days vs 33.80 ± 25.20 days, favoring the AC group. Mirroring our previous results, the mOS difference was maintained in the non-academic setting (mOS for NAC is 44.4m (40.6-54.5) and 37.9m (35.3-40.4) for AC, P = .0004). No difference was found in the academic settings (mOS for NAC 47.4m (43.5-54.0) and 51.9m (47.3-56.0) for AC, P = .4239). The overall survival difference between the NAC and the AC group in this ITT analysis is ~6.5months favoring NAC (44.4 m vs 37.9 m) in the non-academic setting, and there is no difference between the 2 group in the academic settings (the overall survival difference was ~9.3 favoring NAC (47.2m vs 37.9m) in the non-academic setting in IIT analysis a). In the Cox regression model, the hazard ratio in this ITT group is 0.77 (0.62-0.94), P < .0098 in the non-academic setting. The hazard ratio is 1.07 (0.90-1.28), P = .4234 in the academic setting.
As Su and colleague correctly pointed out, the survival benefit was only found in patients treated in non-academic institutions, despite the fact that both types of institutions are subjected to immortal time bias. Therefore, we do not think immortal time bias could account for the difference of survival between the NAC and AC patients treated in the non-academic institutions. The more plausible reason for these difference were discussed in our article.2
Conflict of Interest
Zhonglin Hao reported research funding from Bayer Inc. and consulting/advisory relationships with Pfizer and BeiGene. The other authors indicated no financial relationships.
Contributor Information
Zhonglin Hao, Division of Medical Oncology, Biostatistics and Bioinformatics Shared Resource Facility, Markey Cancer Center, College of Medicine, University of Kentucky, Lexington, KY, USA.
Quan Chen, Division of Medical Oncology, Biostatistics and Bioinformatics Shared Resource Facility, Markey Cancer Center, College of Medicine, University of Kentucky, Lexington, KY, USA.
Bin Huang, Division of Medical Oncology, Biostatistics and Bioinformatics Shared Resource Facility, Markey Cancer Center, College of Medicine, University of Kentucky, Lexington, KY, USA.
References
- 1. Su I-H, Lund JL, Gaber CE, Sanoff HK.. Neoadjuvant versus adjuvant chemotherapy for resectable metastatic colon cancer in non-academic and academic programs. The Oncologist. 2023;28(3):00-00. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Hao Z, Parasramka S, Chen Q, et al. Neoadjuvant versus adjuvant chemotherapy for resectable metastatic colon cancer in non-academia and academic programs. The Oncologist. 2023;28(1):48-58. 10.1093/oncolo/oyac209. [DOI] [PMC free article] [PubMed] [Google Scholar]