This work was presented at AHNS, Chicago, August 2006.
First-year quality of life assessment of an intra-arterial (RADPLAT) versus intravenous chemoradiation phase III trial†
Version of Record online: 28 OCT 2008
Copyright © 2008 Wiley Periodicals, Inc.
Head & Neck
Volume 31, Issue 1, pages 77–84, January 2009
How to Cite
Ackerstaff, A. H., Balm, A. J. M., Rasch, C. R. N., de Boer, J. P., Wiggenraad, R., Rietveld, D. H. F., Gregor, R. T., Kröger, R. and Hilgers, F. J. M. (2009), First-year quality of life assessment of an intra-arterial (RADPLAT) versus intravenous chemoradiation phase III trial. Head Neck, 31: 77–84. doi: 10.1002/hed.20937
- Issue online: 9 DEC 2008
- Version of Record online: 28 OCT 2008
- Manuscript Accepted: 5 JUN 2008
- quality of life;
- intra-arterial infusion;
- IV infusion;
- head and neck cancer;
- advanced stage
We report the results of a multicenter randomized phase III study, assessing quality of life (QOL) in intra-arterial (IA) versus standard intravenous (IV) chemoradiation in advanced head and neck cancer.
Two hundred seven patients with inoperable stage IV disease—152 men and 55 women; mean age, 55 years—were included in this study. The patients were treated with standard radiotherapy with 4 weekly IA or 3 weekly IV cisplatin infusions. The QOL assessments carried out were EORTC-C30, H&N35, and trial-specific questionnaires.
Overall QOL deteriorated in all patients during treatment, is gradually improving over 1 year. IA patients showed significantly less nausea and vomiting at week 7 (p <.001). IV patients were significantly more fatigued (p <.006). At 1 year, no significant difference in tube feeding was found. Voice quality slightly exceeded the pretreatment values at 1 year. Forty-two of 62 employed patients returned to work.
During treatment, significantly fewer problems with nausea and vomiting occurred in IA than in IV patients. Both groups showed improved voicing and oral intake during follow-up, often exceeding pretreatment values at 1 year. © 2008 Wiley Periodicals, Inc. Head Neck, 2009
Quality of life (QOL) issues have become indispensable study outcome parameters when the impact of different treatment modalities in head and neck cancer has to be assessed. Locally advanced head and neck cancer, if treated by surgery followed by radiotherapy, may change the anatomy and may therefore end in problems concerning functionality and disfigurement.1 The areas most commonly affected by this approach are eating, speaking, and, despite advances in reconstruction, facial appearance.1–6 These problems have a serious impact on the psychosocial and overall well being. And, on top of these problems, the prognosis of these patients is rather poor; 5-year survival rates often do not exceed 30%.7
Nowadays, an additional treatment modality, concurrent chemoradiation, is increasingly used for the treatment of locally advanced disease.7–10 A clinical study from our institute, evaluating the effect of intensive targeted chemoradiation (RADPLAT), showed that an initial complete local tumor response could be achieved in 91% of the patients, with 1- and 2-year locoregional control rates of 82% and 69%, respectively.10 QOL assessment, as an integral part of this study, also demonstrated that the impact of an organ preservation protocol such as RADPLAT seems somewhat less severe than found earlier with a pure surgical approach in terms of QOL effects.11
Because of the above-mentioned promising results of targeted chemoradiation (RADPLAT) protocol, a multi-institutional randomized phase III study was conducted to assess whether the more complex intra-arterial (IA) cisplatin infusion indeed yields better results than the standard intravenous (IV) chemoradiation approach. QOL assessment was integrated into this trial to evaluate possible differences between both treatment arms in that respect as well. In this work, we present the first-year QOL results of 207 patients, whereas primary study endpoints, including disease-free and overall survival, locoregional and distant control, incidence of adverse events, and incidence and pattern of recurrence between the 2 treatment arms in the total study population (N = 236) will be presented in a separate work.12
PATIENTS AND METHODS
Between December 1999 and November 2004, 207 of 236 patients in the registered, phase III trial (M99RAD; CKVO 2000–01), were referred to the persons responsible for the QOL part of this randomized, multi-institutional study. The 4 hospitals, participating in the QOL part of the study, were The Netherlands Cancer Institute, Medical Center Haaglanden, Radiotherapy Institute Friesland, all in The Netherlands, and the Waikito Hospital in New Zealand. The internal review boards of the participating hospitals approved the study protocol, and written informed consent was obtained from all patients before entering the study. Twelve of the 29 “missing” patients were treated in a fifth participating center, where the infrastructure for an adequate collection of the QOL data was lacking, and the other 17 “missing” patients were randomly distributed to the other centers.
Before referral to the QOL part of this study, patients were randomly assigned (using the minimization technique stratifying by center, T classification, N classification, and tumor site) either into the IA cisplatin arm (N = 104) or into the IV cisplatin arm (N = 103). In the IA arm, patients received weekly cisplatin 150 mg/m2 as a targeted rapid IA injection with sodium thiosulphate IV rescue in weeks 1 to 4.8, 10 In the IV arm, patients received 100-mg/m2 cisplatin as a 4-hour IV infusion on days 1, 22, and 43. Radiotherapy 70 Gy in 35 fractions was administered over 7 weeks starting concurrently with chemotherapy. Roughly, one fourth of the patient population was treated with intensity-modulated radiotherapy (IMRT).
Patient characteristics, including sites and classifications, are shown in Table 1. All patients had inoperable or functionally inoperable (eg, requiring total glossectomy without or even with total laryngectomy) disease. Forty-six percent of the patients were employed, 32% did not have a job, and 22% of the patients were retired before they started with their treatment. The majority of patients had a history of smoking and (social) alcohol consumption. Just before the start of the treatment, 99 patients (49%) were still smoking (mean, 12 cigarettes per day; range, 1–60) and 84 patients (40%) were still consuming alcohol (mean, 3 U per day; range, 0.15–10 U per day). The remaining patients indicated that they had just quit these habits.
|IA(N = 104)||IV(N = 103)||Total(207)|
|Age (mean)||54||56||55 y|
|T3*||31 (30%)||34 (33%)||65 (31%)|
|T4||73 (70%)||69 (67%)||142 (69%)|
|N0||16 (16%)||21 (20%)||37 (18%)|
|N1||16 (16%)||9 (9%)||25 (12%)|
|N2||61 (59%)||64 (62%)||125 (61%)|
|N3||10 (10%)||10 (10%)||20 (10%)|
|Oral cavity||22 (21%)||18 (17%)||40 (19%)|
|Oropharynx||62 (60%)||67 (65%)||129 (62%)|
|Hypopharynx||20 (19%)||18 (17%)||38 (18%)|
As shown in Table 2, during the first 12 months of follow-up, 60 patients died (32 patients in the IA group and 28 in the IV group), and at the 12-month assessment point, 21 patients could not be interviewed (refusal 2 [1 IA, 1 IV], too ill 4 [2 IA, 2 IV], protocol violation 3 [all IA], extensive salvage surgery 4 [2 IA, 2 IV], and missing 8 [4 IA, 4 IV]). This resulted in 126 patients at 12 months for further detailed analysis (IA arm N = 60, IV arm N = 66).
|Pre||7 wk||3 mo||12 mo|
QOL assessment was performed before treatment, and at 7 weeks (last day of treatment period), 3 months, 1, 2, and 5 years after initiating treatment, using the EORTC QLQ C-30 core questionnaire and the special head and neck module, EORTC H&N 35.13, 14 In addition, to get more detailed information about certain aspects, such as eating and speaking (see below), a structured, validated trial-specific questionnaire (TSQ) was used as well.2, 11
Because the questionnaires are rather extensive, in this analysis, we primarily focused on those aspects most likely to be affected by treatment and most relevant for patients with cancer in the head and neck. The scales of interest derived from the QLQC30 were physical functioning, role functioning, emotional functioning, cognitive functioning, social functioning, fatigue, nausea and vomiting, and dyspnea. From the QLQH&N questionnaire, pain, swallowing, trouble with social eating, dry mouth, sticky saliva, pain killers, and feeding tube were studied. And from the TSQ, the items/scales such as “eating” (diet, mastication, oral transport, and swallowing), speaking (intelligibility, loudness, pitch, and fluency), social contacts, and social anxiety (summated Likert scales) were evaluated. Furthermore, the TSQ includes some items addressing the (pretreatment and posttreatment) smoking and/or alcohol habits, the work situation, and the satisfaction with the pretreatment and posttreatment information c.q. counseling.
All EORTC scale and item scores are linearly transformed to a scale of 0 to 100. A high score for the functional scales and for the overall QOL scale represent a better QOL. In contrast, which is somewhat confusing, for the single items and for the symptom scales, a higher score means a higher level of symptoms and thus a lower QOL. The scores are calculated according to the EORTC QLQ scoring manual.
The data were analyzed using the statistical package Splus 6.2 and SPSS 12.0 for Windows. In case of missing responses for 1 or more items of the multi-item EORTC scales, a “half-scale” solution was adopted, in which at least half of the responses to the items of a given scale are available. The mean score for the remaining items of the scale then replaces the missing item responses for that particular patient. If more than half of the item responses for a given multi-item scale were missing, then the scale score for that patient was treated as missing.
For the scales of interest, the maximum change from pretreatment levels was calculated for each patient. Nonparametric Wilcoxon rank sum test was conducted to determine if the distributions for the 2 arms significantly differed. The resulting p values were adjusted for multiple testing using the Sidak technique.
For 2 scales, fatigue and nausea and vomiting, further multivariate analysis was conducted. For the former, this was to determine if there was a significant difference between trial arms in the duration of fatigue experienced by patients. For the latter, this was to determine if the significant difference between the levels of nausea and vomiting in the 2 trial arms was independent of other factors.
In the case of the fatigue scale, a linear mixed model was constructed in which trial arm, time since randomization, pretreatment fatigue levels, sex, age at randomization, time off-study, and tumor classification were taken as fixed effects, and a patient variable was taken as a random effect. The time covariate was modeled using a natural cubic spline with 3 degrees of freedom. Interaction terms were investigated, and a Wald F test was used to determine the significance of excluding the trial arm variable from the model. It is noted that the scale date is ordinal rather than numeric; however, the distribution of values for the fatigue scale was such that the linear mixed effects model assumptions were not violated.
For the nausea and vomiting scale, an ordinal proportional odds model was constructed, using the same predictor variables as in the fatigue model. Proportionality and ordinality assumptions were checked and interaction terms were examined.
Correlations were assessed by Pearson's correlation coefficient. A 2-tailed p value <.05 was considered statistically significant.
The fact that only a subgroup of patients (207 of 236) participated in the QOL assessment part of this study did not result in distribution problems, as can be seen in Table 1; there were no differences in age, sex, tumor site, or classification in both treatment groups.
Figure 1 presents the distributions for 13 scales of the maximum change from pretreatment to 12 months by trial arm for all participating patients (the IA arm is depicted on top and the IV arm below that), including the (raw) p values. Especially, scales such as physical functioning, role functioning, and also social functioning are negatively influenced by the treatment. Also, an increase in problems such as dry mouth and sticky saliva can be observed. The only symptom scale showing a statistically significant difference between the 2 groups (irrespective of sex, age at randomization, time off-study, and tumor classification) is the nausea and vomiting scale (raw p value <.001, adjusted p value = .001). Twice as many patients in the IV group than in the IA group complained of these symptoms at 7 weeks (p <.001; Table 3). At 3 months, these complaints/differences have almost disappeared. Fifty-nine percent (N = 56) of the IV patients reported weight loss during treatment compared to 49% (N = 43) of the IA patients, but this difference was statistically not significant (p = .23).
|IA (N = 88)||IV (N = 95)|
|Nausea (quite a bit/very much)||26 (30%)||50 (53%)*|
|Vomiting (quite a bit/very much)||13 (15%)||33 (35%)*|
Another interesting finding is the difference in fatigue level between the 2 arms and pretreatment and 7 weeks and 3 months (p <.006). Patients in the IV group dying before the second or third interview had a significantly higher level of fatigue, already pretreatment, than patients completing the 1-year follow-up and patients in the IA arm (Figure 2).
Focusing on the patients who were able to complete the 1-year follow-up (N = 126), irrespective of treatment arm, many of the EORTC C30 functioning scales appeared to deteriorate at the end of the 7-week-treatment period followed by a gradual improvement up to 12 months, sometimes exceeding the pretreatment levels (Table 4). Also, the items/scales of the H&N35 representing problems specifically related to H&N cancer treatment showed an increase in problems at 7 weeks, followed by a decrease over time. Especially, pain and speech problems, considerably increased at week 7, showed a clear improvement at 12 months compared with the pretreatment data. As expected, a Wilcox test showed that patients who indicated they used painkillers, especially during the treatment period, experienced more pain than those who did not take them (p <.001; no differences between both treatment arms). The majority (96%) of the patients had no problems with their appearance 12 months after initiating treatment.
|Baseline||7 wk||3 mo||12 mo|
|Head and neck pain||28.5/33.0||47.7/41.0||28.1/26.0||12.1/13.0|
There was an increase over time in problems such as dry mouth and sticky saliva. Xerostomia (quite a bit to very much) was reported by 35 patients (52%) in the IV group and 36 patients (60%) in the IA group, whereas sticky saliva was reported by 37 patients (56%) in the IV group and by 26 patients (46%) in the IA group, but also this difference as not statistically significant (p = .7).
An overview of patients who needed tube feeding during the 12-month assessment period is shown in Table 5. Pretreatment 31 patients in the IA and 30 in the IV group needed tube feeding, whereas, at month 12, 10 of 60 IA patients (16%) still had tube feeding, compared with 16 of 66 (24%) in the IV group (not statistically significant; p = .3). Based on the TSQ, it appeared that the remaining 100 patients were able to return to a more or less normal diet at month 12. At that moment, 61% of the patients reported that the sense of taste was good or excellent, whereas 39% of the patients indicated that the taste was still poor to moderate.
|7 wk (end of treatment)||79/88||78/95|
Voice quality (intelligibility—face-to-face, as well as on the telephone—loudness, pitch, and fluency; also assessed with the TSQ) deteriorated during the 7-week treatment period, but then gradually improved over time, slightly exceeding the pretreatment values at month 12. As a result, 83% of the patients reported that they had reasonably good understandable voice at month 12, with no statistically significant differences between both groups. However, the question “is your voice like it used to be?” was answered affirmatively by only 33% (same percentage in IA as in IV group) of the patients.
For both the groups, social contacts, visiting, or receiving friends/family decreased during and immediately following treatment and went back to the usual pretreatment routine at 1-year follow-up. Of the 62 patients, who were employed before treatment, 42 were able to return to their job (gradually from a couple of hours per day for some to their regular intensity for most of them) within 12 months (22 in IA group and 20 in IV group). At month 12, approximately 50% of the patients smoked (mean 9 cigarettes per day) and consumed alcohol (mean 2.5 U per day). Many of them, who had not stopped already before the treatment (49% and 40%, respectively), did not quit this habit during or immediately after treatment.
Ninety-three percent of all participating patients were quite satisfied with the pretreatment information. No differences over time or differences between groups could be found. They also indicated that they understood quite well what the effect of the treatment (temporarily) might have on functional aspects such as eating and speaking and their appearance. However, it appeared that the posttreatment information was given less regularly. Many patients expressed the need for more follow-up information about aspects such as the duration of tube feeding and/or when the sense of taste would be back to normal.
In summary, our study results show that most well-known treatment-related QOL problems do not significantly differ between treatment arms, except for nausea, vomiting, and fatigue, which are significantly less severe in the IA arm than in the IV arm.
Because any chemoradiation treatment, either through IA or IV administration, poses a heavy burden on patients, it is expected that all participating patients (irrespective of treatment arm) showed a clear deterioration in most QOL aspects at the end of the 7-week-treatment period followed by a gradual improvement up to 12 months. Obvious examples are the changes in physical, role, and social functioning scales. The negative change in physical functioning is inherent in the 7-week treatment, whereas a decrease in role functioning and also social functioning is primarily due to the fact that many patients are hospitalized at the end of the treatment.
In general, it can be said that the groups of patients did not differ much with respect to QOL issues. The only statistically significant differences between the 2 groups concerned nausea and vomiting and feelings of fatigue (pretreatment, 3 months). Twice as many patients in the IV group than in the IA group had complaints of nausea and vomiting at the 7-week assessment point. However, these feelings of nausea and vomiting gradually disappeared over the next months. Concerning the feelings of fatigue, it appeared that the patients in the IV arm who died before the second or third assessment point were more troubled by the feelings of fatigue, also already pretreatment, than patients in the IA arm. This finding confirms in part some of the results mentioned in our RADPLAT IA feasibility study.11 In that study, patients who died or had to undergo extensive salvage surgery within 1 year following RADPLAT IA treatment, often already pretreatment, had less positive scores on functional and physical well-being scales and on several H&N scales of the FACT questionnaire than patients benefiting from the RADPLAT treatment. In general, it might be suggested that a “good” pretreatment condition may influence or predict the outcome of this treatment in a positive way.
Feelings of pain in the head and neck area, clearly present pretreatment and increased during treatment, diminished during the 12-month follow-up. As a result, the need for (nonnarcotic) painkillers decreased as well. An exception on these improvements over time was the development of xerostomia (involving a decrease in the amount of saliva and a thicker, stickier consistency), a common encountered consequence of radiotherapy. In this study, approximately half of the patients had xerostomia problems, whereas the literature mentions percentages between 50% and 77%.15–18
This somewhat more favorable result might be due to the fact that approximately 1 fourth of the patients were treated by means of IMRT. Also, in this randomized chemoradiation study, there was a frequent need for tube feeding, especially during the first 3 months. It is a commonly seen consequence of the treatment that patients lose weight due to trouble in swallowing and eating problems.19, 20 And, despite the tube feeding, more patients in the IV group reported weight loss during treatment in comparison with the IA group. Feelings of nausea and vomiting, more present in the IV group, were probably of influence as well. At month 12, approximately 20% of the patients (N = 126) still needed tube feeding. This percentage is in accordance with our previous RADPLAT IA study, in which 19% of the patients still needed tube feeding at the 1-year follow-up. Also, researchers from the University of Chicago reported that following chemoradiation treatment, 18% to 20% of the patients were unable to take anything orally at 12 months.4 The remaining 100 patients were able to return to a more or less normal oral diet. Some patients remarked that eating meat presented some problems, especially if the meat was somewhat leathery. Also, because of xerostomia, many patients have to drink (more than they were used to) during meals to get the food down. For two-third of the patients, taste returned to normal within 12 months. However, some patients remarked that especially spicy food tasted sometimes somewhat peculiar and that it was different before the chemoradiation treatment. From the literature4 and from personal observation (based on incidental patient remarks), some continued improvement with regard to aspects of taste is possible, even after 12 months.
For approximately 80% of the patients (same percentage as in the RADPLAT feasibility study11), the voice was reasonable to good. However, only 33% of the patients reported that their voice/speaking was like it used to be. The remaining patients mentioned that they got tired of speaking sooner, resulting in a weaker voice or a somewhat hoarse voice especially at the end of the day; and/or that to keep on talking requires more effort, also often due to dry mouth/lack of saliva.
It is a positive finding that two thirds of the patients who had a job before starting treatment were able to return to that job within 12 months, also very similar to our previous phase II IA study.11
Fifty percent of the patients continued or returned to their smoking and alcohol habits, but one has to keep in mind that many already had (temporarily) quit these habits before the treatment (51% and 60%, respectively). These percentages were also found in the RADPLAT pilot study as well as in the study assessing the consequences of composite resections.2, 11 Because all patients are nowadays informed about the risks of smoking and drinking, many patients hasten to say that they drink and smoke less than before their treatment. However, it has to be said that it can be assessed correctly by means of a questionnaire whether a patient smokes or drinks alcohol, but when it comes to the number of cigarettes or the amount of alcohol units per day you may assume that those answers are probably not reliable.
As a result of our earlier studies and to meet the need for more follow-up information as expressed by some patients in this study as well an extra outpatient clinic visit, approximately half-year following treatment was initiated by the nursing staff. During this visit, patients could indicate for which subjects more information was needed. Most of the time, the nursing staff could resolve these questions themselves, but sometimes they had to refer the patient to the appropriate persons/departments.
This 1-year follow-up analysis reveals that the differences in QOL issues between the IA and the IV group are quite limited and maybe somewhat less than expected in view of the potential differences in toxicity between treatment arms. As expected, however, it could be shown that patients in the IA group had significantly less problems with nausea and vomiting at the end of the treatment than patients in the IV group. Also, with respect to the feelings of fatigue, the IA group “scores better”: these feelings are more prominent in the IV group, especially in those patients who died before the 3-month assessment point. Problems with voicing and oral intake, clearly present at the second assessment point in both groups, improved during the 12-month follow-up, often exceeding pretreatment values. The fact that two thirds of the patients were able to return to their job within the 12-month follow-up period emphasizes the extent of recovery. Xerostomia remains a residual complication inherent to the nature of radiotherapy.
Although the IA treatment thus appears to cause less early toxicity, the decision whether this advantage outweighs the advantage of the less complicated IV chemoradiation mainly will depend on the clinical outcomes with regard to locoregional control and disease-free survival. And for this we have to wait for the final analysis of the completed trial.
The authors greatly appreciate the expert statistical assistance of Andrew Vincent (Department of Biometrics), and also thank Angela Goossens (Medical Centre Haaglanden) and Dick Alberts (Radiotherapy Institute Friesland) for their help in collecting the QOL data in their respective hospitals.
- 1Psychological aspects of major head and neck surgery. Facial Plast Surg 1995; 2: 91–98., , , , .
- 5Rehabilitation after head and neck surgery. In: JonesAS, PhillipsDE, HilgersFJM, editors. Diseases of the head and neck, nose and throat. London: Edward Arnold; 1998. pp 435–448., .
- 12Intra-arterial versus intravenous chemoradiation for advanced head and neck cancer, early results of a multi-institutional trial. Proceedings of the 2006 annual meeting and research workshop on the biology, prevention and treatment of head and neck cancer, Chicago, August 17–20, 2006 (abstract S016, p 84)., , , et al.