Proton versus photon therapy for oropharyngeal cancer - Authors' reply.

# Proton versus photon therapy for oropharyngeal cancer – Authors' reply *The Lancet 2026* Correspondence grade 3 xerostomia. In the present overall survival difference should be Authors’ reply study, the xerostomia and dysphagia attenuated in the intention-to-treat We thank David J Sher for the grade 3 numbers are highly collinear versus per-protocol comparison, which compliments on our phase 3 trial (61 [45%] of 136 patients experienced is not demonstrated. An as-treated of intensity-modulated (photon) xerostomia and 67 [49%] patients overall survival analysis with crossover radiation therapy (IMRT) versus experienced dysphagia in the IMRT group specifics is needed. Second, intensity-modulated proton therapy group vs 52 [33%] of 160 patients and 22 patients in the proton group whose (IMPT) for oropharyngeal cancer.1 54 [34%] patients in the IMPT group)— insurance was denied were immediately A key finding from the trial was therefore, it can be hypothesised that lost to follow-up and presumably that proton therapy de-intensified xerostomia was incorrectly coded as censored at t=0. This creates a prima chemoradiation therapy, reducing grade 3 due to concurrent grade 3 facie case of informative censoring; the rates of high-grade toxicities and dysphagia. An alternative hypothesis a poor insurance product is conceivably gastrostomy tube dependency to is that unstimulated saliva output was related to the overall survival endpoint. 40·2% in the IMRT group and 26·8% used for the grading; those data should Third, the authors did not control type in the IMPT group. Sher expressed be presented to support the results I error for multiple hypothesis tests concern that the reported rate of if this was the case. Moreover, the and, since the a priori hypothesis was severe xerostomia (45% IMPT and hypothetical improvement from IMPT non-inferiority, it is possible type I error 33% IMRT) was higher than that in the depends on improved doses to organs for the overall survival superiority test RTOG 1016 trial,2 where 243 (61·5%) at risk, and it is important to share this occurred. of 395 patients given IMRT and dosimetry to explain the presented To save the causal connection cisplatin required a gastrotomy tube. results. In the case of xerostomia, between protons and overall survival, Enteral feeding during chemo- comparative parotid, submandibular, Frank and colleagues suggested that radiation therapy reflects a progressive, and oral cavity dosimetry would help the adverse impact on immunity in the multifactorial process that includes inform this discussion. photon group might have altered the dysgeusia, oral mucositis, xerostomia, choice or efficacy of salvage therapies. nausea, vomiting, dysphagia, and I declare no competing interests. However, the authors do not show weight loss. Therefore, one or more David J Sher data on the salvage therapies received of these severe toxicities likely david.sher@utsouthwestern.edu nor comparative blood counts in contributed to the need for enteral Department of Radiation Oncology, The University progressors from the two groups. support. The grade 2–3 xerostomia rate of Texas Southwestern Medical Center, Dallas, The above concerns, coupled with in RTOG 1016 (53·5%) might reflect TX 75390, USA negative data from the TORPEdO trial,2 under-reporting of grade 3 toxicities 1 Frank SJ, Busse PM, Lee JJ, et al. Proton versus photon radiotherapy for patients with call into question the conclusion that contributing to 243 (61·5%) patients oropharyngeal cancer in the USA: proton therapy is the primary driver of needing enteral support during a multicentre, randomised, open-label, non-inferiority phase 3 trial. Lancet 2026; the overall survival improvement seen. chemoradiation therapy. 407: 174–84. Forthcoming sialometry and We declare no competing interests. 2 Gillison ML, Trotti AM, Harris J, et al. dosimetric data are expected to provide Radiotherapy plus cetuximab or cisplatin in *Sean M McBride, Nadeem Riaz, human papillomavirus-positive oropharyngeal a more comprehensive understanding Eric J Sherman, C Jillian Tsai, cancer (NRG Oncology RTOG 1016): of how IMPT reduces xerostomia a randomised, multicentre, non-inferiority Loren K Mell trial. Lancet 2019; 393: 40–50. during chemoradiation therapy. mcbrides@mskcc.org We disagree with Sean M McBride Department of Radiation Oncology, Memorial Sloan and colleagues and stand by Kettering Cancer Center, New York, NY 10065, USA Trials of local therapies where (SMM, NR); Department of Medicine, Division of our finding regarding improved progression-free survival is equivalent Solid Tumor Oncology, Memorial Sloan Kettering overall survival with IMPT. IMPT is but overall survival is improved raise Cancer Center, New York, NY, USA (EJS); Department a paradigmatic example of a local of Radiation Medicine, Princess Margaret Cancer the spectre of confounding. The therapy that can achieve comparable Centre, Toronto, Canada (CJT); Department of overall survival improvement seen Radiation Medicine and Applied Sciences, Division disease control rates to conventional in the phase 3 trial by Steven J Frank of Radiation Oncology, San Diego, USA (LKM) IMRT while improving overall survival. and colleagues1 of proton therapy 1 Frank SJ, Busse PM, Lee JJ, et al. Proton versus McBride and colleagues suggest that photon radiotherapy for patients with for oropharyngeal cancer is likely an if proton therapy improved overall oropharyngeal cancer in the USA: example of this issue. First, if proton a multicentre, randomised, open-label, survival, attenuation of effect would non-inferiority phase 3 trial. Lancet 2026; therapy truly increased survival— be seen when comparing intention- 407: 174–84. given the crossover from photon 2 Thompson DJ, Price JM, Tyler M, et al. Proton to-treat with per-protocol analyses. to proton and barring extreme bias beam therapy for oropharyngeal cancer With the bidirectional crossover, (TORPEdO): a phase 3, randomised controlled in the crossover population—the there was attenuation (hazard ratio trial. Lancet 2026; 407: 1259–75. Correspondence 0·55 per-protocol vs 0·58 intention- feeding at 1 year, disease control, and to-treat) with a reduction in hazard early overall survival estimates. The of death from 45% to 42% favouring 5-year overall survival estimates in our IMPT. The robustness of the overall trial were 90·9% for IMPT and 81·0% survival estimates across the for IMRT; however, in the TORPEdO intention-to-treat, per-protocol, and trial, the 5-year data remain to be as-treated analyses (not reported) seen. Therefore, we can only conclude also demonstrated consistent that the absence of evidence is not improvements in overall survival with evidence of absence. IMPT. SJF declares grant support for the trial from Hitachi Second, McBride and colleagues paid to the University of Texas MD Anderson Cancer suggested that 22 patients in the Center (UTMDACC); support from the UTMDACC (P30CA016672), Massachusetts General Hospital proton group whose insurance was (U19CA021239), National Institutes of Health denied were immediately lost to (R03CA188162 and P01CA285249), and National follow-up and presumably censored Association of Proton Therapy paid to UTMDACC; consulting and honoraria fees from Ion Beam at t=0, which was not the case. In Applications and Boston Scientific (not related to total, 17 patients receiving IMRT and this work); ownership interest, patents, royalty fees, 26 patients receiving IMPT withdrew and fees for the Scientific Advisory Committee for C4 Imaging (not related to this work); is a Board after randomisation, and their exact Member for the National Association of Proton censoring times were applied; the two Therapy and National Comprehensive Cancer early deaths were counted as events Network; and is Chair of the Proton Therapy Oncology Group Head and Neck Subcommittee. in the intention-to-treat analysis. RLF reports royalty fees from Elsevier and Bionix Survival follow-up was preserved for (both not related to this work). All other authors all crossover patients. We acknowledge declare no competing interests. that the type of insurance policy could *Steven J Frank, Paul M Busse, reflect factors influencing overall J Jack Lee, Robert L Foote, on behalf of survival, but we no found clear the University of Texas MD Anderson evidence that insurance denial was Cancer Center Clinical Trial Consortium associated with lower socioeconomic sjfrank@mdanderson.org status. Arguably, the protective effect Department of Radiation Oncology (SJF) and of IMPT would have been even greater Department of Biostatistics (JJL), University of Texas if there were a shortage of recources MD Anderson Cancer Center, Houston, TX 77030, USA; Department of Radiation Oncology, for managing the side-effects of Massachusetts General Hospital, Boston, MA, USA chemoradiation therapy. (PMB); Department of Radiation Oncology, Mayo Third, McBride and colleagues Clinic, Rochester, MN, USA (RLF) suggested that without adjustment 1 Frank SJ, Busse PM, Lee JJ, et al. Proton versus photon radiotherapy for patients with for multiple hypotheses tests, the oropharyngeal cancer in the USA: type I error was not controlled, a multicentre, randomised, open-label, which is incorrect—our study was non-inferiority phase 3 trial. Lancet 2026; 407: 174–84. designed as a non-inferiority trial 2 Gillison ML, Trotti AM, Harris J, et al. with progression-free survival as the Radiotherapy plus cetuximab or cisplatin in human papillomavirus-positive oropharyngeal primary endpoint. The overall survival cancer (NRG Oncology RTOG 1016): analysis was based on the secondary a randomised, multicentre, non-inferiority endpoint, which does not warrant trial. Lancet 2019; 393: 40–50. 3 Thompson DJ, Price JM, Tyler M, et al. Proton adjustment for multiple tests. beam therapy for oropharyngeal cancer Lastly, although the TORPEdO trial3 (TORPEdO): a phase 3, randomised controlled trial. Lancet 2026; 407: 1259–75. had a different design, fewer patients, and shorter follow-up, outcomes in the two trials were strikingly similar— ie, IMPT reduced high-grade toxicities and gastrostomy tube dependence at the end of treatment relative to IMRT. In both trials, IMPT and IMRT achieved similar rates of enteral 1918 --- [PDF原文](https://sci-net.xyz/storage/7932541/07b24de2f77c516df5d28d4822d001b6dc832adf5fb042f30a05507f823eeeef/Proton-versus-photon-therapy-for-oropharyngeal-cancer-Authors-reply.pdf) DOI: 10.1016/S0140-6736(26)00793-2