The RADIANCE-NMA project, led by researchers searching PubMed, Embase, Cochrane CENTRAL and ClinicalTrials.gov from inception to 26 June 2026, identified 2,830 records and ultimately included 67 randomised trials, of which 42 reporting overall survival formed the primary network. The trials spanned 1977 to 2025 and more than 32 countries, involving major cooperative groups such as RTOG, EORTC, GORTEC and DAHANCA. Most were Phase III studies, and the population was restricted to adults with locally advanced squamous cell carcinoma of the oral cavity, oropharynx, hypopharynx or larynx, with nasopharyngeal cancer excluded. The researchers used a frequentist random-effects network meta-analysis with radiotherapy alone as the reference treatment, and validated their rankings with a parallel Bayesian analysis.

The headline result concerns which treatments actually prolong life compared with radiotherapy alone. Four strategies achieved statistically significant overall survival benefits: cisplatin plus 5-fluorouracil showed a hazard ratio of 0.35 (95 percent confidence interval 0.18 to 0.66), nimotuzumab plus weekly cisplatin reached 0.48 (0.29 to 0.81), weekly cisplatin achieved 0.57 (0.44 to 0.74), and three-weekly cisplatin delivered 0.79 (0.65 to 0.95). Paclitaxel plus cisplatin showed a borderline result at 0.42, while cetuximab plus radiotherapy, long considered the standard alternative for platinum-ineligible patients, approached but did not reach significance at 0.81 (0.64 to 1.03).

The ranking analysis placed cisplatin plus 5-fluorouracil at the top with a P-score of 94.8 percent, followed by paclitaxel plus cisplatin, nimotuzumab plus weekly cisplatin, weekly cisplatin and intra-arterial cisplatin. Cisplatin three-weekly, the long-standing standard, ranked only 15th, and cetuximab ranked 16th. At the bottom of the hierarchy sat misonidazole, panitumumab and erythropoietin, the latter showing a non-significant unfavourable point estimate that echoes earlier concerns about erythropoietin receptor stimulation on tumour cells. Radiotherapy alone ranked 20th of 24 treatments. Concordance between the frequentist P-scores and Bayesian SUCRA values was excellent, with a mean absolute difference of just 0.4 percentage points, giving the authors strong methodological confidence in the ordering.

One of the most consequential findings concerns cisplatin scheduling. The network estimate favoured weekly over three-weekly cisplatin with a hazard ratio of 0.73 (0.58 to 0.93), a 27 percent reduction in the hazard of death. Yet the authors urge caution: the underlying head-to-head trials, including JCOG1008, the Noronha long-term update and two small Indian Phase IIb studies, were all designed as non-inferiority or exploratory comparisons, and none was powered to demonstrate weekly superiority. The network estimate gains apparent statistical significance by integrating direct and indirect evidence across the broader treatment graph, a methodological feature rather than a head-to-head superiority demonstration. Notably, the Noronha update showed superior locoregional control with three-weekly dosing, and the cumulative dose hypothesis, in which weekly schedules achieve higher completion rates exceeding 200 milligrams per square metre, may partly reconcile the divergent endpoints.

The analysis also delivered a sobering verdict on cetuximab-based chemoradiotherapy. Compared with three-weekly cisplatin, the network estimate showed no significant difference (hazard ratio 1.01, 0.81 to 1.27), but the authors explicitly state that equivalence cannot be formally claimed because the confidence interval is wider than typical equivalence margins. Individual head-to-head trials consistently favoured cisplatin: De-ESCALaTE HPV, RTOG 1016 and TROG 12.01 all demonstrated inferiority of cetuximab in HPV-positive oropharyngeal cancer, and the recent ARTSCAN III trial reinforced cisplatin superiority with a hazard ratio of 1.66 against cetuximab. Interestingly, biomarker subgroup data suggested marked cetuximab benefit in p16-positive tumours but none in p16-negative disease, hinting at a potential niche role requiring prospective validation.

Immunotherapy combinations fared poorly. KEYNOTE-412 testing pembrolizumab plus chemoradiotherapy showed non-significant event-free survival, JAVELIN Head and Neck 100 with avelumab was terminated for futility, and the COMPARE trial with durvalumab was negative. The network analysis confirmed that neither pembrolizumab plus weekly cisplatin nor pembrolizumab plus concurrent chemoradiotherapy achieved significant overall survival benefit, though treatment-related mortality was lower with immunotherapy at 1.0 percent compared with 2.2 percent for platinum regimens and 3.7 percent for EGFR-targeted agents. The authors suggest radiation-induced lymphodepletion as a plausible mechanism of failure and call for low-dose or attenuated-frequency immunotherapy schedules and sequencing strategies in future trials.

Xevinapant, an inhibitor of apoptosis protein antagonist that had generated excitement after a Phase II trial showed a hazard ratio of 0.47, disappointed in Phase III. The TrilynX trial did not meet its primary endpoint, and with Phase II and Phase III data pooled into a single node, the combined hazard ratio versus radiotherapy alone was 0.76 (0.48 to 1.20), no longer statistically significant. The authors describe this as confirmation that early-phase signals require Phase III validation before being considered practice-changing. On the toxicity side, severe mucositis rates were highest for carboplatin plus 5-fluorouracil at 71.0 percent and nimotuzumab plus weekly cisplatin at 66.7 percent, while cetuximab plus radiotherapy produced the highest severe dermatitis rate at 56.1 percent. Cisplatin plus 5-fluorouracil occupied the most favourable efficacy-toxicity position, combining the lowest hazard ratio with moderate mucositis of 38 percent.

Perhaps the most important caveat concerns time. Meta-regression showed that enrolment year significantly modified treatment effects, with later trials showing smaller incremental benefits. Stratifying at a 2005 cutoff, three-weekly cisplatin versus radiotherapy alone yielded a hazard ratio of 0.81 (0.68 to 0.96) in the pre-2005 era but only 1.03 (0.45 to 2.36) after 2005, a statistically significant interaction. The authors attribute this attenuation to the transition from two-dimensional or conformal radiotherapy to intensity-modulated radiotherapy, which has narrowed the incremental benefit of adding a radiosensitiser. They recommend that present-day treatment decisions be informed primarily by post-2005 evidence. Oropharyngeal cancer proportion also significantly modified effects, consistent with HPV-positive tumours having favourable prognosis regardless of treatment, diluting apparent sensitiser benefit.

The study has acknowledged limitations: it relied on aggregate study-level data rather than individual patient data, precluding HPV-stratified analyses; risk of bias assessment found 79 percent of assessed studies had some concerns, largely from open-label designs; and top-ranked treatments rest on limited evidence with wide confidence intervals. The network also excluded altered-fractionation-only arms, so it cannot claim cisplatin-based strategies outperform radiation intensification alone. Nevertheless, the practical message is clear. Cisplatin-based chemoradiotherapy remains the preferred option for patients who can tolerate platinum therapy, weekly cisplatin stands as an acceptable alternative, cetuximab is best reserved for the cisplatin-ineligible, and the search for the next breakthrough radiosensitiser, whether nanoparticle platforms such as NBTXR3 or DNA damage response inhibitors, must continue through adequately powered trials.

Subject of Research: Network meta-analysis of radiosensitising strategies in locally advanced head and neck squamous cell carcinoma

Article Title: Radiosensitising strategies in locally advanced head and neck cancer: a systematic review and network meta-analysis of randomised controlled trials (RADIANCE-NMA)

Article References: MP, S., Dange, P., Rao, K. N., Harrington, K. J., Vermorken, J. B., Forastiere, A. A., Mendenhall, W. M., Ferlito, A., Agaimy, A., Araújo, A. L. D., Asarkar, A., Bal, M., Beitler, J. J., Bradford, C., Bradley, P. J., Bradová, M., de Bree, E., de Bree, R., Chow, J. C. H., … Zidar, N. (2026). Radiosensitising strategies in locally advanced head and neck cancer: a systematic review and network meta-analysis of randomised controlled trials (RADIANCE-NMA). eClinicalMedicine, 101, Article 104263. https://doi.org/10.1016/j.eclinm.2026.104263

Image Credits: AI Generated

DOI: 10.1016/j.eclinm.2026.104263

Keywords: head and neck cancer, radiotherapy, cisplatin, network meta-analysis, cetuximab, immunotherapy, xevinapant, nimorazole, HPV, radiosensitisers, systematic review, eClinicalMedicine