Patients and Treatment
Between October 2012 and June 2019 a total of 470 patients were registered in 97 German centers. 79 patients were excluded as they violated at least one eligibility criteria (figure 1). The most frequent eligibility criteria violation was a missing combination of cetuximab with radiotherapy or platinum-based chemotherapy. The remaining 391 patients were included in the TAS. Seventy-six patients provided no evaluable pair of questionnaires before and during therapy, and thus in the mTAS 315 patients were evaluable. Clinical characteristics of the 391 TAS patients are given in table 1. 198 patients presented with local recurrence only (50.6%), 119 patients had distant metastases only (30%) and 74 patients (19%) had local relapse and distant metastases. 77 patients with an ECOG score of ≥2 were included (20%) and 124 patients had a Charlson comorbidity score greater than one (32%). Treatment consisted of cetuximab plus radiotherapy in 78 patients (20%) and cetuximab plus chemotherapy in 309 patients (79.0%), 4 patients received both (1%). The chemotherapy was cisplatin based in 174 patients (56%) and carboplatin based in 139 patients (44%). 264 patients had received prior surgery (68%) and 323 patients prior radiotherapy (83%).
Response to Treatment
The mean follow-up time was 8.6 months (range: 0 – 33.8). The ORR in the entire cohort was 33% (95%CI: 28.8 – 38.1) and DCR was 56% (95%CI: 51.3 – 61.1). In the subgroup chemotherapy-cetuximab the ORR was 32% and in the subgroup radiotherapy-cetuximab 39%. In addition, Kaplan-Meier analyses of OS and PFS were performed in the entire cohort (TAS cohort) (supplementary figure S2). The median PFS was 5.5 months (95%CI: 4.8 – 6.0) and the median OS was 9.5 months (95%CI: 8.5 – 10.9).
Baseline symptom burden
Baseline symptoms of the 315 evaluable patients (mTAS) are given in Figure 2. The mean overall VAS score before treatment was 35.4, slightly worse than the mean score for pain with 31.3. Most severe symptoms at baseline were swallowing problems with solid food (mean 57.7), followed by speech problems (mean 40.5), and restriction of physical activities (mean 38.3). The self-assessed mean actual overall heath state was 46.1 and thus worse than most of the single symptoms. In addition, baseline symptoms were analyzed separately in patients with locoregional recurrence only (without distant metastases) and in patients with distant metastases (or both). Patients with locoregional recurrence had worse baseline swallowing fuction of solid/ mashed food and liquids and more speech problems compared to patients with distant metastases (supplementary figure S3).
Correlation of treatment response and tumor symptoms
Changes in the patients’ symptom burden are studied for responders (CR/PR) versus non-responders (SD, PD, NA). All changes are displayed for the three time points: The “first post-baseline” assessment compares the first assessment during treatment with the baseline values. The “best post-baseline” assessment compares the best post baseline values of any questionnaire during treatment with the baseline values. The “end of treatment” assessment compares the values at treatment termination with the baseline values. Negative values indicate improved symptoms and positive values deteriorated symptoms.
The change of overall VAS score from baseline was significantly better in responders compared to non-responders at the first post-baseline assessment (LSM responders -2.13 vs. non-responders +1.15, p=0.0476) (figure 3). This effect became stronger, when the best post-baseline assessment was chosen (LSM responders -7.82 vs. non-responders -1.97, p=0.0005). At end of therapy the mean overall tumor symptom score returned to baseline in responders and deteriorated in non-responders (LSM responders +0.78 vs. non-responders +6.99, p=0.0088).
An additional analysis of changes of the overall VAS score in patients with locoregional recurrence only (without distant metastases) and in patients with distant metastases (or both) was performed. Changes of the overall VAS score at the three time points “first post-baseline assessment”, “best post-baseline assessment” and “assessment at treatment end” were similar to the changes in the entire cohort and did not differ in patients with locoregional recurrence and distant metastases (supplementary figure S4).
The results of the ten single symptom sub-VAS scores are presented in figure 4. In the swallowing assessment, especially solid food was a problem for the patients. At the best post-baseline assessment swallowing of solid food improved significantly stronger in responders (LSM -16.67 vs. non-responders -5.06, p=0.0016) (figure 4A). For swallowing mashed or liquid food also significant differences in favor of responders were observed at the best post-baseline and the end of therapy assessment (figure 4B, C). Larger differences were seen for the symptom pain (figure 4D). At the best post-baseline assessment the mean pain score has improved considerably more in responders than in non-responders (LSM: responders -16.37 vs. non-responders -8.89, p=0.0011). Similar to swallowing problems, also restriction of smell or taste both were significantly better in responders when the best post-baseline assessment was compared (figure 4 E, F). Also speech problems were a main impairment of patients. Speech problems significantly improved more in responders at the best post-baseline assessment (responders -13.25 vs. non-responders -4.60, p=0.0027) and remained better until end of treatment (responders -3.38 vs. non-responders +5.78, p=0.0154) (figure 4G). Responders and non-responders reported no significant differences in breathing problems (figure 4H). Responders also evaluated their physical activity and current health state better than non-responders in the best post-baseline assessment (figure 4 I, J).
Time from treatment initiation is also an important factor. Changes from baseline of the overall VAS score and the single symptoms are reported in monthly intervals (supplementary figure S5). Whereas the overall VAS score continuously improves during the first three months in responders, it clearly worsens in non-responders especially at the third month and later. Especially swallowing function of mashed food and liquids slowly improves in responders, whereas it dramatically worsens in non-responders after three months. Physical activity remains stable in responders, whereas it worsens in non-responders already after two months and later.
Association between baseline factors and OS
In the univariate cox regression analysis to study potentially prognostic factors on OS, older age (especially those between 66 to 75 years), a Charlson score of 0, lower ECOG scores, female sex, and a less severe overall VAS score were associated with lower mortality risk considering all variables with an effect p-value of < 0.2 (table 2). However, alcohol consumption, body weight, type of therapy (RT only, CT only, RCT), duration since initial diagnosis, location of primary tumor (oropharynx, hypopharynx, larynx, mouth/ lip/ tongue, multiple locations, other), type of relapse (loco-regional only, any distant metastases) and smoking status (non-smoker, former smoker, current smoker) were not associated at a p-level of 0.2.
In the multivariable analysis only the overall VAS score remained a prognostic factor for overall survival, with hazard increase of 12% per 10 points increment for the overall VAS score at baseline. (hazard ratio: 1.12 per 10 points in VAS, 95% CI 1.05-1.20, p=0.0009) (table 2).