Rationale

Long term results of randomised controlled trials (RCT) testing dietary and/or physical interventions in cancer patients are scarce. In most studies, the objective is behavioural change between inclusion and the end of the training period (that is, from a few weeks to several months) and long-term evaluations usually do not exceed 1 year post-intervention (Zhu et al, 2016). Also, patients’ adherence to recommendations often decline significantly after the supervised intervention is completed (Courneya et al, 2007, 2009; Trinh et al, 2014; Greenlee et al, 2016). The PACThe trial is a 2-week physical/nutritional intervention in hydrothermal centre (SPA Sana-Per-Aquam), whose main objective was an increase of health-related quality of life (QoL) 1 year post-intervention, with follow-up over 5 years. PACThe protocol and primary outcomes have been described in previous articles (Kwiatkowski et al, 2013; Mourgues et al, 2014). Long-term results are now available.

Patients and methods

PACThe was a prospective multicenter RCT testing the impact on QoL of a single 2-week group intervention (7–12 patients at a time). Other endpoints included sleep, anxiety/depression, weight control and some biological parameters. Intervention included daily group supervised physical training, dietary education, physiotherapy and psychological support. Three hydrothermal cities from central France participated in the trial: Vichy, Le-Mont-Dore and Châtel-Guyon. Selection criteria were non-metastatic breast cancer patients with invasive tumours, aged from 18 to 75 years, in complete remission after treatment of their disease (including chemotherapy).

Anthropometric measures were performed by nutritionists (weight, waist circumference and body composition by impedanceometry) at inclusion and 6, 12, 18, 24 and 36 months post-inclusion. Patients answered questionnaires at the same periods, plus at year 4 and 5 after inclusion: SF36 for QoL (Ware and Sherbourne, 1992; Ware et al, 1994; Richard et al, 2000), HAD hospital anxiety and depression (Zigmond and Snaith, 1983; Lépine et al, 1985) and Ricci and Gagnon questionnaire for physical/sedentary activities (Walger, 2009).

Patient consent was obtained before the study. The trial was performed in compliance with the Helsinki declaration and approved by AFSSAPS (French Agency for Sanitary Security of Health Products), the French National Committee controlling personal computerised data (CNIL), and the regional Ethics Committee (CPP Sud-Est-6 in March 2008). It was registered in ClinicalTrials.gov (NCT01563588).

Statistics

Impact of the intervention at different time points is described with effect-sizes (ES=difference between mean values by allocation group divided by the common standard deviation) and their 95% confidence intervals (95% CI). QoL scores are standardised with a 0–100 scale. All tests are two-sided and P-values0.05 are considered significant. SEM software (Kwiatkowski et al, 2000) was used to perform the statistics.

Results

Flowchart of accrual and response rates per period

The response rate at 5 years was 65% in the SPA group and 58% in the control group (60% overall). Anthropometric data were obtained at 2 years in 82 and 82% of participants, respectively, and at 3 years in 43 and 46% of participants (Figure 1). Patient characteristics, reported in our 2013 article, are not duplicated here. Median follow-up of participants was 65 months (13–96).

Figure 1
figure 1

Inclusion flowchart and data count at each time-point. Qol, quality of life; diet, nutritional+anthropometric data.

Evolution of patients’ quality of life over 5 years

Overall patients’ QoL (all dimensions together) was significantly improved by the SPA intervention (Figure 2A), the widest difference between curves occurring just after the intervention (measured at 6 months):

Figure 2
figure 2

Evolution of SF36 quality of life scores (ranking from 0=worst to 100=best) according to allocation group (A) QoL global score (B) Mental and Physical sub-scores (error bars correspond to 95% confidence interval).

The effect-size of the intervention on QoL over the whole follow-up period was 0.33 (0.23; 0.43). Effect-sizes at various time points were: 6 months=0.63 (0.37; 0.89), P=0.000003; 1 year=0.29 (0.03; 0.55), P=0.032; 18 months=0.29 (0.02; 0.56), P=0.037; 2 years=0.27 (−0.01; 0.56), P=0.06; 3 years=0.28 (−0.02; 0.58), P=0.064; 4 years=0.41 (0.02; 0.81), P=0.039 and 5 years=0.45 (0.11; 0.80), P=0.0093.

To limit possible attrition bias and curtail the person-effect, this comparison was made using the QoL variation from baseline (inclusion) in percents. Similar shapes of curves per group were obtained. The overall effect-size was however smaller: 0.19 (0.08; 0.30) (P=0.0009)

Considering the two main dimensions of the SF36 questionnaire (Ware et al, 1994), the impact of the intervention was significant (P<10−7) for both the mental and the physical sub-scores (difference between curves; Figure 2B). The correlation between both scores was stronger among participants from the SPA group (r=0.85) than among the control group (r=0.80, r-difference P=0.0012). SF36 physical and mental scores were highly correlated to the level of physical activity (resp. r=0.29 and r=0.27, P<10−7).

Anxiety and depression evolution over 5 years

Intervention had a limited impact on anxiety, except at 6 months (P=0.025). Depressive symptoms were more influenced by the intervention at 6 months P=0.000014, effect-size=−0.46 (−0.73;−0.19) and at 1 year −0.34 (−0.61;−0.08). However, gains after 1 year were not significant (−0.28effect-size−0.12 till 5 years).

Anthropometric measures and activity/sedentarity

Over the 3-year follow-up, 20% participants from both groups had weight increase of>5%. On average, weight variation from baseline was −1.1% in the SPA group (95% CI=(−1.7; −0.6)) vs 0.7% (0.2; 1.2) in the control group (P=0.0000016). Figure 3 exhibits variations of the weight and abdominal circumference from baseline at each time: for the weight, the overall effect-size was −0.42 (−0.55; −0.29) (p≈10−5). This was similar if patients were stratified according to BMI at inclusion: for BMI<25 kg m2, the probability associated to the difference was 0.00045 for BMI>25 kg m2, P=0.0021. Overall, the impact of the intervention on weight control did not last more than 2 years.

Figure 3
figure 3

Weight and waist-girth variations by allocation group over 3 years (=% differences between measures at inclusion and those at each time points) Error bars correspond to 95% confidence interval).

The impact of the intervention on the reduction of abdominal adiposity (measured using waist-girth variation from baseline) was stronger: on average over 3 years, −2.0% (−2.5; −1.4) in the SPA group vs 0.7% (−0.1; 1.5) in the control group (P<10−7).

Overall, weight variations significantly correlated with the level of physical activity (P=0.000002), which influenced weight variation by ≈5% (r=−0.17). The SPA intervention interacted with this correlation (P=0.002): correlation between activity and weight decrease was significant in the SPA group (r=−0.24, P=0.000001) but not in the control group (r=−0.05, P=0.29). The multivariate analysis of parameters influencing abdominal circumference over time selected only three significant items: allocation group (P=2.6 × 10−7), caloric input (P=0.0000043) and physical activity (P=0.0026). In particular, menopausal status and time after intervention were not significant.

Discussion

Because of its simple design (a 2-week multidisciplinary intervention in SPA) and the promising middle term results (Kwiatkowski et al, 2013), PACThe trial appeared to be an efficient strategy to help breast cancer survivors improve their QoL. The effect-size of the intervention on QoL was moderate at 6 months and small at 1 year. PACThe performed better at 6 months than standard protocols: in a large meta-analysis (Duijts et al, 2011), effect-size of physical activity on QoL was 0.30 (0.12; 0.48). Adherence and gains usually fade with time (Courneya et al, 2007, 2009; Greenlee et al, 2016). This 5-year update shows that improvements obtained post-intervention are remarkably durable. PACThe intervention seems to provide a benefit that propels patients to another QoL level, unattainable under usual circumstances.

However, the 60% 5-year response rate in our study limits the reliability of our estimates and the strength of our conclusion: an attrition bias cannot be excluded even though analysis of individual variations from baseline remained significant (P=0.0009).

Depressive symptoms are one of the most frequent side-effects of cancer, even after complete remission (Massie, 2004; Reich et al, 2008). The gains provided by the intervention at 6 months and at 1 year were higher than those reported in two meta-analyses (Duijts et al, 2011; Craft et al, 2012). This advantage lasted till 5 years, but not significantly.

Weight control in our study did not exceed 2 years, after significant differences at 6 months and 1 year. Interestingly, the intervention was more efficient on abdominal adiposity than on weight, with more lasting and significant differences in abdominal circumference reduction. Considering the particular negative impact of central adiposity on breast cancer prognosis (Ibrahim and Al-Homaidh, 2011; Sun et al, 2015), our intervention appears to be a good strategy to limit this risk. The risk reduction of premature mortality and cardio-metabolic diseases is also concerned (Nichols et al, 2009; Després, 2012). The PACThe intervention ameliorated participants’ perceived self-efficacy and their positive behavioural beliefs on the adherence to long-term behavioural changes (Aparicio-Ting et al, 2015).

Conclusion

Overall, our intervention demonstrated many advantages: PACThe only necessitated a 2-week participation for patients. This combination of physiotherapy, physical exercises, nutrition and group support appeared to induce a real motivation shift in patients which permitted a durable amelioration of QoL that correlated with more rapid resumption of professional and occupational activities. Moreover, SPA costs were totally compensated by the reduction of professional absenteeism and medical fees following chemotherapy (Mourgues et al, 2014). Such a strategy should be investigated to other cancer locations where remission rates are high and QoL a priority target.