Prognostic impact of weight loss during radiation therapy in patients with head and neck cancer: A systematic review

Introduction

Head and neck cancer (HNC) rank as the seventh most common malignancy worldwide and include a heterogeneous group of tumours originating in the paranasal sinuses, nasopharynx, oral cavity, pharynx (naso-, oro- and hypopharynx) and larynx (Alterio et al., 2019; Mody et al., 2021). Radiotherapy (RT), with or without chemotherapy (CTX) or surgery, plays a central role in curative treatment, particularly in advanced stages (Alterio et al., 2019).

Despite therapeutic advances, patients undergoing RT or chemoradiotherapy often experience severe side effects such as mucositis, dysphagia and dermatitis, which frequently lead to weight loss (WL) and malnutrition (Ghadjar et al., 2015; Han et al., 2021; Löser et al., 2024). While pretreatment nutritional status is recognized as a prognostic factor, the impact of WL occurring during RT on overall survival (OS) and disease-free survival (DFS) remains unclear (Finger et al., 2024). WL in HNC is not only driven by reduced intake due to mucositis and dysphagia but can also reflect systemic inflammation and tumour-related catabolism. These mechanisms may reduce physiologic reserve, impair treatment tolerance and affect survival. This systematic review aims to synthesize the available evidence on the prognostic value of WL alongside RT in patients with HNC.

Patients and methods

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines; all eligibility criteria, data extraction steps and outcomes were predefined to minimize selection and reporting bias. The review protocol was not registered (e.g. PROSPERO). A structured search of PubMed/MEDLINE and the Cochrane Library was conducted in November 2024, using Medical Subject Headings (MeSH) and Title/Abstract terms related to head and neck cancer, radiotherapy, nutrition, weight loss or body mass index (BMI), and survival/prognosis. The full electronic search syntax is provided in Appendix. Embase and Scopus were not searched; this is a limitation of the review.

Studies published from 2012 onward were included to ensure comparability with contemporary standards of HNC care. Around this time, intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) became routine, replacing older 3D-conformal techniques and substantially changing dose distribution, toxicity profiles and nutritional impact. Because IMRT/VMAT allow better sparing of swallowing-related structures and salivary glands, they may reduce dysphagia- and xerostomia-driven reductions in oral intake compared with older 3D techniques. Therefore, restricting inclusion to studies from 2012 onward improves clinical comparability of treatment toxicity and its nutritional consequences. In parallel, concurrent cisplatin-based chemoradiotherapy was the established standard of care, and structured nutritional screening and support protocols were increasingly implemented in clinical practice. Eligibility criteria included: (i) histologically confirmed HNC; (ii) treatment with RT or chemoradiotherapy; (iii) human studies; (iv) quantitative assessment of WL, defined either in kilograms (kg) or via BMI change; (v) reporting of survival outcomes. Exclusion criteria comprised case reports, reviews, letters, editorials and duplicates.

Definitions and terminology

To ensure consistency across studies, the following terminology was applied throughout the review:

WL: any documented reduction in body weight or BMI relative to baseline, expressed either as percentage of initial body weight or as ΔBMI in kg/m².

Critical WL: a loss >5% within 8 weeks or >7.5% within 12 weeks, according to the international consensus by White et al. (2012)

Pretreatment WL: WL occurring within 6 months before the start of RT.

WL during RT: WL documented between the first and last day of RT or chemoradiotherapy.

Posttreatment WL: WL measured after completion of RT, up to 6 months post-therapy.

Where individual studies used alternative definitions, data were standardized to the extent possible for descriptive comparison. Data extraction focused on study characteristics, treatment details, WL metrics and survival outcomes. We considered limited subgroup pooling (e.g., pretreatment WL >10% and ΔBMI >1 kg/m²), but meta-analysis remained inappropriate due to substantial heterogeneity and incomplete reporting of effect estimates. Definitions of intratreatment WL ranged from ≥5% to ≥10%, and assessment windows differed (end of RT, 2 months post-RT, or up to 5 months post-RT), alongside variation in endpoints (OS, DSS/cancer-specific survival (CSS), DFS, progression-free survival (PFS), distant metastasis-free survival (DMFS)/locoregional recurrence-free survival (LRRFS)). Because effect estimates and time points were not sufficiently comparable, statistical heterogeneity metrics (e.g., I²) were not calculated. A PRISMA flow diagram is detailed in Figure 1.

OPEN IN VIEWER

Figure 1.

PRISMA flow diagram of literature search and study selection; WL: weight loss.

The methodological quality of all included studies was assessed independently by two reviewers using the Newcastle–Ottawa Scale (NOS) for cohort studies. When details were unclear (e.g., handling of missing data, exact follow-up time), conservative assumptions were applied to avoid overestimation of study quality. Inter-rater agreement between the two reviewers was κ = 0.84, indicating strong concordance, disagreements were resolved by consensus. A summary of the NOS ratings is provided in Supplementary Table S1.

Quantitative pooling of data was deemed inappropriate because of marked heterogeneity among studies regarding tumour subsites, treatment modalities, weight-loss definitions and survival endpoints. A narrative synthesis was therefore conducted. However, where studies shared comparable definitions (e.g., pretreatment WL > 10%), results were examined descriptively to identify consistent patterns. Because no quantitative pooling was performed, studies were not statistically weighted by sample size. In the narrative synthesis, we placed greater emphasis on larger cohorts and higher-quality studies when deriving overall conclusions.

Results

Following a comprehensive review of the extant literature, 164 articles were identified as potentially relevant. However, subsequent to the application of the inclusion criteria, eight articles were determined to meet all the specified criteria and thus were included in the final analysis. The eight articles incorporated both retrospective and prospective designs. The included studies analysed various tumour subsites (e.g., nasopharynx, oropharynx, oral cavity), RT techniques (e.g., IMRT and VMAT), WL definitions and endpoints. This provides a broad perspective on the prognostic relevance of WL in modern HNC care. Table 1 provides a summary of key study characteristics and outcomes.

OPEN IN VIEWER

Table 1.

Overview of included studies on the prognostic impact of weight loss (WL) and BMI in head and neck cancer.

Author (year)	Tumour entity/RT setting	Outcomes	WL definition (timing/threshold)	N (population details)	Key findings
Langius et al. (2013)	Mixed HNC (RT ± surgery ± CTX)	5-year OS; DSS	Pre-WL > 10% (≤6 mo); WL during RT > 5% ≤ 8 wk or >7.5% ≤ 12 wk	1340	Pre-WL > 10% → worse OS (HR 1.7; P = 0.002) and DSS (HR 2.1; P = 0.007). WL during RT > 5% → worse DSS (HR 1.7; P = 0.004) but not OS
Ghadjar et al. (2015)	Locally advanced HNC (hyperfractionated RT ± cisplatin)	5-year OS, DMFS, LRRFS, CSS	Pre-WL ≥ 10% (≤6 mo); WL during RT ≥ 10%	224	Pre-WL ≥ 10% → worse OS (HR 3.2; P < 0.001), LRRFS (HR 2.5; P < 0.001), DMFS (HR 3.1; P < 0.001), CSS (HR 2.8; P = 0.002). WL during RT→ no impact.
Han et al. (2021)	Mixed HNC (definitive CRT 71%, post-op RT 29%)	Median OS; 5-year OS; CSS	WL during RT ≥ 5.8% (dichotomized at median)	843	High WL (≥5.8%) → better OS (HR 0.75; P = 0.01) and CSS (HR 0.71; P = 0.01). Interpretation likely affected by confounders.
Moon et al. (2016)	HNSCC (definitive CRT + high-dose cisplatin)	OS; CSS; PFS	Pre-BMI < 18.5 kg/m2; WL during RT (ΔBMI pre-vs 2 month post-RT)	153	Low Pre-BMI < 18.5 → poor OS (HR 15.2; P < 0.001) and CSS (HR 20.1; P < 0.001). WL during RT → no association.
Lin et al. (2015)	nasopharyngeal carcinoma (IMRT ± CTX)	3-year OS; DSS	WL during RT ≥ 5%; Pre-BMI < 23 kg/m2	238	WL during RT ≥ 5% → no impact on OS (P = 0.087) or DSS (P = 0.232). Pre-BMI < 23 → no effect.
Lin et al. (2022)	nasopharyngeal carcinoma (VMAT ± CTX)	5-year OS; DFS; D-FFR; ΔBMI	During RT-ΔBMI > 1 kg/m2; Pre-BMI categories	498	ΔBMI > 1 kg/m2 → lower 5-yr OS (73.0% vs 88.0%; P < 0.001), DFS (65.1% vs 80.4%; P < 0.001), D-FFR (78.5% vs 90.9%; P < 0.001). Independent predictor of poor outcome.
Cho et al. (2013)	Oral / oropharyngeal SCC (mixed modalities)	DFS	Post-WL ≥ 10% (≤6 month post-RT)	226	Post-WL ≥ 10% → reduced DFS (P = 0.036); independent predictor of poor DFS. OS not reported.
Ottosson et al. (2014)	oropharyngeal carcinoma (ARTSCAN trial)	5-year OS	WL during RT ≥ 10% (start RT → 5 mo); Pre-BMI > 25 kg/m2	357	WL during RT ≥ 10% → no impact on 5-yr OS (P = 0.708). Pre-BMI > 25 kg/m2→ better OS (P < 0.05). WL correlated with irradiated volume.

WL: weight loss; Pre-WL: pretreatment weight loss within 6 months before RT; WL during RT: weight loss assessed during RT or in the early post-RT period, as defined in the individual studies; Post-WL: weight loss within 6 months after RT; ΔBMI: change in body mass index (kg/m²) from baseline to the study-specific assessment time point (end of RT or early post-RT, as reported); BMI: body mass index; OS: overall survival; HR: hazard ratio; DSS: disease-specific survival; DFS: disease-free survival; CSS: cancer-specific survival; PFS: progression free survival; LRRFS: locoregional recurrence-free survival; DMFS: distant metastasis-free survival; D-FFR: distant failure-free rate; RT: radiotherapy; CRT: chemoradiotherapy; CTX: chemotherapy; HNC: head and neck cancer; HNSCC: head and neck squamous cell carcinoma; VMAT: volumetric modulated arc therapy.

An overview of the timing and direction of associations is illustrated in Figure 2.

OPEN IN VIEWER

Figure 2.

Timeline of weight-loss assessment and corresponding prognostic impact across eight included studies. In the two studies assessing pretreatment WL ≥10% within 6 months before RT, this was consistently associated with worse survival outcomes (OS/DSS) (↓). Baseline BMI showed prognostic relevance, with low BMI associated with adverse OS and higher BMI associated with better OS. WL during RT/early post-RT showed variable effects across studies: most reported no association (↔), two reported adverse associations (one for DSS and one for OS/DFS/D-FFR using ΔBMI) (↓), and one reported an apparent favourable association (↑; likely confounding). Posttreatment WL within 6 months after RT was associated with reduced DFS (↓). Arrows indicate direction of association (↓ adverse, ↔ no association, ↑ apparent favourable).

Langius et al. conducted a prospective analysis including 1340 patients with HNC treated with curative intent using RT, either alone or in combination with surgery ± CTX. The study investigated the prognostic significance of WL before and during RT on 5-year OS and disease-specific survival (DSS). Critical WL during RT was defined, in accordance with the international consensus statement by White et al. (2012), as a body weight reduction of >5% from the start of RT until week 8, or >7.5% until week 12. This condition was observed in 57% of patients. The mean WL during RT was 4.1 kg (±4.7), corresponding to 5.4% (±6.1%) of baseline body weight. Unadjusted 5-year OS and DSS were significantly lower in patients with critical WL during RT compared to those without (OS: 62% vs. 70%, P = 0.01; DSS: 82% vs. 89%, P = 0.001). In multivariable analysis, critical WL remained significantly associated with impaired DSS (hazard ratio [HR] 1.7, 95% CI [1.2–2.4], P = 0.004), but not with OS (HR 1.1, 95% CI [0.9–1.4], P = 0.295). Additionally, pre-treatment WL showed a dose-dependent prognostic effect: while 70% of patients had no WL before RT, 16% had WL ≤5%, 9% had 5–10% and 5% had >10% WL. Corresponding 5-year OS rates declined with increasing WL (71%, 59%, 47% and 42%, respectively; P < 0.001), and the DSS rates followed a similar trend (86%, 86%, 81% and 71%; P < 0.001). In adjusted models, pre-treatment WL >10% was independently associated with worse OS (HR 1.7, 95% CI [1.2–2.5], P = 0.002) and DSS (HR 2.1, 95% CI [1.2–3.5], P = 0.007) (Langius et al., 2013).

Ghadjar et al. analysed 224 patients with locally advanced head and neck squamous cell carcinoma (HNSCC) enrolled in the randomized phase III SAKK 10/94 trial. Patients were treated with hyperfractionated RT, either alone or in combination with cisplatin-based CTX. The study evaluated the prognostic value of WL both before and during treatment. WL was measured retrospectively for the 6 months prior to therapy and prospectively during treatment. A threshold of ≥10% WL prior to treatment initiation was identified as clinically relevant. In multivariable analyses, pretreatment WL ≥10% was significantly associated with worse outcomes across all survival endpoints, including OS (HR 3.2, 95% CI [2.0–5.3], P < 0.001), LRRFS (HR 2.5, 95% CI [1.4–4.5], P < 0.001), DMFS (HR 3.1, 95% CI [1.6–6.1], P < 0.001) and CSS (HR 2.8, 95% CI [1.5–5.4], P = 0.002). In contrast, WL during treatment – although common (occurring in 51% of patients) – was not significantly associated with any of the analysed survival endpoints in either univariable or multivariable analysis. The authors emphasize the need for early nutritional intervention and vigilant monitoring prior to the initiation of treatment (Ghadjar et al., 2015).

Han et al. conducted a retrospective single-centre analysis of 843 patients with HNC who received RT with curative intent, either definitively (71%) or postoperatively (29%), between 2003 and 2017. WL during treatment was assessed as a continuous variable, with the cohort dichotomized at the median WL of 5.8%. Patients with high WL (≥5.8%) showed significantly improved median OS (39.2 vs. 36.7 months; P = 0.047) and 5-year OS (54.9% vs. 48.8%; P = 0.047) compared to patients with lower WL. Similarly, 5-year CSS was higher in the high WL group (64.0% vs. 58.2%; P = 0.036). In multivariate Cox regression, high WL was independently associated with improved OS (HR 0.75, 95% CI [0.61–0.93], P = 0.01) and CSS (HR 0.71, 95% CI [0.55–0.93], P = 0.01) (Han et al., 2021).

Moon et al. conducted a prospective study involving 153 patients with HNSCC undergoing definitive chemoradiotherapy. Body weight and BMI were recorded within 2 weeks before and 2 months after treatment. Although a significant reduction in weight and BMI was observed during therapy, neither WL nor post-treatment BMI change showed a significant association with PFS, CSS, or OS. In contrast, low pretreatment BMI (<18.5 kg/m²) was independently associated with worse survival outcomes, including CSS (HR 20.12, 95% CI [4.82–83.96], P < 0.001) and OS (HR 15.23, 95% CI [4.06–57.11], P < 0.001). Additional predictors of adverse outcome included pretreatment hypoalbuminemia and elevated neutrophil-to-lymphocyte ratio (NLR), while changes in body weight alone were not found to be prognostically relevant (Moon et al., 2016).

Yu-Hsuan Lin et al. performed a retrospective study on 238 patients with nasopharyngeal carcinoma (NPC) treated with IMRT. The prognostic value of pretreatment BMI and WL during therapy was evaluated. Patients were stratified by BMI (<23 vs. ≥23 kg/m²) and WL (≥5% vs. <5%), with 63% of the cohort experiencing WL ≥5% during treatment. However, neither parameter was significantly associated with OS or DSS in univariate or multivariate analysis. Specifically, WL ≥5% showed no negative impact on 3-year OS (P = 0.087) or DSS (P = 0.232), and BMI stratification showed no prognostic value (OS: P = 0.672; DSS: P = 0.341). Subgroup analyses also revealed no interaction between baseline BMI and the magnitude of WL. Sensitivity analyses using different cutoff values for BMI and WL confirmed these findings (Lin et al., 2015).

Xiang Lin et al. conducted a retrospective single-centre study of 498 patients with NPC treated with VMAT. The prognostic relevance of pretreatment BMI and treatment-related BMI change (ΔBMI) was assessed. Pretreatment BMI showed no statistically significant association with OS, DFS, distant failure-free survival (D-FFR), or locoregional control. In contrast, a decrease in BMI >1 kg/m² during treatment was significantly associated with inferior oncologic outcomes. Patients with ΔBMI >1 kg/m² had significantly lower 5-year OS (73.0% vs. 88.0%, P < 0.001), DFS (65.1% vs. 80.4%, P < 0.001) and D-FFR (78.5% vs. 90.9%, P < 0.001) compared to those with ΔBMI ≤1 kg/m². Multivariate analysis confirmed ΔBMI >1 kg/m² as an independent predictor of poor outcome, whereas pretreatment BMI alone showed no prognostic significance. The authors concluded that ΔBMI is a simple and clinically accessible marker to identify patients at increased risk and emphasized the need for timely nutritional intervention in this population (Lin et al., 2022).

Cho et al. conducted a retrospective study analysing 226 patients with previously untreated, operable squamous cell carcinoma (SCC) of the oral cavity (n = 123) or oropharynx (n = 103), treated at a single centre between 2005 and 2010. Patients underwent heterogeneous treatment modalities including surgery, RT, CTX, or combinations thereof. Body weight was measured at several time points, and the first day of treatment was defined as the baseline. Posttreatment WL ≥10% was observed in 24.2% of patients and was attributed to both tumour-related and treatment-induced factors such as RT, disease recurrence and the presence of ≥3 metastatic lymph nodes. Multivariate logistic regression identified RT, posttreatment recurrence and multiple nodal metastases as independent predictors of WL ≥10%. Additionally, significant WL was associated with worse DFS (P = 0.036). In multivariate analysis, WL ≥10% remained an independent predictor of poor DFS. OS was not reported (Cho et al., 2013).

Ottosson et al. conducted a retrospective cohort study using data from 357 patients with oropharyngeal carcinoma enrolled in the ARTSCAN trial. The study investigated the prognostic significance of WL and BMI in relation to 5-year OS, as well as the association between WL and irradiated volume. WL was defined as the percentage weight change from the start of RT to 5 months post-treatment and dichotomized at a 10% threshold. WL ≥10% occurred in 68% of evaluable patients but was not significantly associated with 5-year OS (77.0% vs. 74.1%; P = 0.708). In contrast, baseline BMI >25 kg/m² was significantly associated with improved survival outcomes. Patients classified as overweight or obese had a 5-year OS of 83.2%, compared to 58.8% and 56.7% in the underweight and normal-weight groups, respectively. In multivariate Cox regression, BMI remained an independent prognostic factor (HR 3.78 for underweight, 95% CI [1.46–9.75]; HR 2.57 for normal weight, 95% CI [1.43–4.62]). Moreover, irradiated volume (TV64.6 Gy and TV43.7 Gy) was significantly correlated with increased WL, highlighting the importance of radiation dose distribution in nutritional outcomes (Ottosson et al., 2014).

Across the two largest high-quality cohorts, pretreatment WL ≥10% showed a consistent association with worse survival, with reported hazard ratios ranging from HR 1.7 to HR 3.2 for OS and from HR 2.1 to HR 2.8 for DSS/CSS. A detailed evidence map is provided in Supplementary Table S2, an overview matrix summarizing prognostic associations by WL timing and outcome is provided in Supplementary Table S3.

Discussion

This systematic review summarizes eight studies assessing the prognostic significance of WL before, during and after RT in HNC. Across studies, pretreatment WL >10% consistently predicted poorer overall and disease-specific survival, whereas the impact of WL during or after RT was less uniform. Methodological heterogeneity and inconsistent WL definitions likely explain the divergent results.

Conclusion

This systematic review highlights heterogeneous evidence on WL as a prognostic marker in HNC. While pretreatment WL is consistently associated with worse outcomes across multiple studies, the prognostic significance of WL during and after RT remains variable. The timing and context of WL appear to influence its clinical relevance, with early nutritional depletion posing greater prognostic concern than treatment-related changes alone. Future research should focus on harmonizing definitions, improving nutritional risk assessment, and evaluating structured interventions to support patients with or at risk of critical WL. Integrated nutritional care should be considered an essential component of multidisciplinary oncologic management. Standardized screening at defined time points could support early dietetic referral and help identify patients who may benefit from intensified supportive care during RT.

Supplemental Material

Supplemental Material