Effect of Chocolate on Older Cancer Patients in Palliative Care: A Randomised Controlled Study

Josiane C. Vettori Ribeirão Preto Medical School University of São Paulo Luanda G. da-Silva Ribeirão Preto Medical School University of São Paulo Karina Pfrimer Ribeirão Preto Medical School University of São Paulo Alceu A, Jordão-Junior Ribeirão Preto Medical School University of São Paulo Paulo Louzada-Junior Ribeirão Preto Medical School University of São Paulo Julio C. Moriguti Ribeirão Preto Medical School University of São Paulo Eduardo Ferriolli Ribeirão Preto Medical School University of São Paulo Nereida K.C. Lima (  nereida@fmrp.usp.br ) Ribeirão Preto Medical School University of São Paulo

IG1 patients were instructed to consume 25g of chocolate containing 55% cocoa daily for 4 weeks, while IG2 consumed 25g of white chocolate. CG was instructed not to consume chocolate. During the study the investigators did not interfere with the habitual food consumption of the patients. None of the volunteers habitually consumed chocolate before the study. Chocolate was supplied in 5 portions of 5g per day, for a total of 140 tablets.
The chocolate containing 55% cocoa provided a daily amount of 1337 mg polyphenols/ml GAE/patient [16]. IG1 received per day: 126 kcal; 12 g carbohydrates, 1.5 g proteins and 8.8 g total fats, while IG2: received 136 kcal; 14 g carbohydrates, 1.4 g proteins, and 8.3 g total fats. The patients recorded daily on a card the amount of chocolate consumed.
Primary otcome (nutritional status, including food consumption, anthropometry, and body composition) and secondary outcome measures (oxidative stress, in ammatory activity, and QL) were analyzed initially and after 4 weeks: -General and health characteristics: sociodemographic and health status data.
-Nutritional status: Mini Nutritional Assessment (MNA), a method used for the geriatric population [17] and validated for the Brazilian population [18].
-Food consumption: 24-hour Diet Recall (24HR) and Food Frequency Questionnaire (FFQ). The FFQ was elaborated based on a food list, calibrated [19] and validated for older adults [20], applied only at the end of the study and used to assess the habitual diet consumed during the last 6 months. Nutrient consumption was estimated as: frequency of consumption x portion size x nutritional composition [20]. Food consumed was converted to grams and calculated with the Virtual Nutri Plus software updated with the data of the Brazilian Table of Food Composition [21]. The results obtained were compared to recommended intake of macro-and micronutrients for the age range [22].
The intake of total polyphenols was quantitated using the Phenol-Explorer databank, version 3.0 [23].
-Body composition: determined by the deuterium oxide method after an 8-hour overnight fast. In the morning, each volunteer received 1 ml/kg deuterium oxide (99.9% deuterium oxide, Cambridge Isotope, USA) diluted to 7%, followed by 50 ml natural water for full ingestion of deuterium and mouth washing. Saliva samples were collected before and three hours after intake of the dose. The deuterium enrichment of the samples was determined by isotope ratio mass spectrometry (IRMS, Europa Scienti c Hydra System, Cheshhire, UK) after equilbration with 100% hydrogen by the platinum-alumina catalyzer.
-Routine clinical laboratory tests: blood count, albumin, total proteins, sodium, potassium, and calcium ion.
-In ammatory activity: serum levels of interleukin 6 (IL-6) were determined by ELISA with high sensitivity R&D Systems kits (Minneapolis, MN, USA). C-reactive protein was determined by the latex immunoturbidimetric assay.
-Presence of DNA damage: immunoassay with the DNA/RNA Oxidative Damage EIA Kit (Cayman Chemical) for the detection of all three oxidized quanine species based on 8-hydroxy-2'-deoxyguanosine (8-OHdG) levels.
-Quality of life: application of the instrument of the European Organization for the Research and Treatment of Cancer (EORTC) -QLQ-C30 Questionnaire[28], with 30 questions including scales of overall health status, symptoms and function, with scores of 0 to 100. The higher these scores, the better the QL. High scores on the symptoms scale indicate a poorer QL (Authorization of the EORTC Quality of Life Group).
Data were analyzed statistically using the SAS Statistical Software, version 9.3 (SAS Institute, Inc. Cary, NC, USA) and the R Core Team (2016). Data were submitted to descriptive analysis, categorical variables were analyzed by the chi-square test, and a mixed effects linear regression model was used for group comparison at each time point and and for comparison between time points in each group, adjusted for sex and age, with the level of signi cance set at < 0.05.

Results
-Sociodemographic and clinical characterization of the sample Mean patient age was 67.6 ± 5.7 years (range: 60-83 years) and mean KPS was 88.0 ± 10.9%. Median time since cancer diagnosis was 43.5 months, while median time since the diagnosis of locally advanced or metastatic cancer was 11 months.
Mean chocolate consumption was 136 ± 8.3 tablets of 5 grams each for IG1 and of 135.8 ± 8.8 tablets for IG2. The sociodemographic and clinical characteristics of the patients are listed in Table 1, with no signi cant difference (p > 0.05) between groups. -Characterization of food intake and nutritional status Estimated data of current and habitual macro-and micronutrient intake are presented in Table 2. At the beginning of the study, the energy and protein intake of more than half the volunteers (n = 25; 57.4% and n = 25; 57.4%, respectively) was below the daily recommendations. Mean daily calorie intake was 19.48 ± 4.20 kcal/kg current weight/day and mean protein intake was 0.66 ± 0.22 g /kg current weight /day.
Calorie consumption (Kcal/kg weight) was lower in CG after the intervention period (p = 0.01; 95% con dence interval (CI): +0.5 to + 3.9). Group comparison also showed that calorie consumption was lower in CG than in IG1 both at the beginning (p = 0.04; CI= -11.9 to -0.4) and at the end (p < 0.01; CI= -15.1 to -3.5) of the study.
Except for sodium, the intake of bers, calcium and of all vitamins analyzed (A, B6, B12, C and E) was below recommended levels.
Regarding nutritional status, initially 43.5% of the patients were at risk (n = 15; 32.6%) or were malnourished (n = 5; 10.9%) according to the MAN tool. However, at the beginning of the study, IG1 patients had a lower score at screening (p < 0.01; CI = + 0.5 to + 2.8) and during nutritional assessment (p = 0.04; CI = + 0.1 to + 4.1) with the MAN tool compared to CG. IG1 patients showed an increase in the screening (p < 0.01; CI= -2.2 to -0.4) and nutritional assessment (p = 0.04; CI = + 0.1 to -4.1) scores at the end of the study (Table 3).  At the beginning, CG had a higher BMI than IG1 (p = 0.04; CI = + 0.1 to + 6.0) and IG2 (p = 0.04; CI = + 0.2 to + 6.1), which continued to be higher than that of IG1 (p = 0.03; CI = + 0.4 to + 6.3) at the end of the study (Table 3).
-Laboratory exams and QL  Regarding the antioxidant capacity, GSH levels were lower in IG1 than in CG (p = 0.02; CI = + 0.2 to + 2.9) and IG2 (p < 0.01; CI= -3.4 to -0.7) at the end of the study, with an increase in IG2 (p = 0.04; CI= -1.6 to -0.02). In contrast, vitamin C levels were lower in the intervention groups than in CG at the beginning (p < 0.01; CI = 0.07 to 0.2 between IG1 and CG; CI = 0.08 to 0.2 between IG2 and CG) and at the end of the study (p < 0.01; CI = 0.1 to 0.3 between IG1 and CG; CI = 0.1 to 0.3 between IG2 and CG).
Lipid peroxidation, determined according to MDA levels, was reduced in IG2 (p = 0.02; CI = + 0.7 to + 9.1) from the beginning to the end of the study and IL-6 levels were higher in IG1 (p = 0.03; CI = + 12.9 to + 219.3) than in IG2 at the end of the study.
The QL of IG1 patients (Table 5) improved in terms of functionality, with a higher score for the functional domain (p = 0.03; CI= -13.3 to -0.7), the role functioning subdomain (p < 0.01; CI= -36.4 to -6.3), and the social subdomain (p < 0.001; CI= -28.8 to -4.8). There was no deleterious effect that could be attributed to the consumption of dark or white chocolate, such as nausea, vomiting, diarrhea, or epigastric pain.

Discussion
The present study, conducted on older cancer patients in palliative care with preserved functionality, demonstrated bene ts in terms of improved nutritional status and QL in the group ingesting chocolate with a higher percentage of cocoa. IG1 showed an increased estimated polyphenol intake at the end of the intervention compared to CG and IG2. Several studies that used the values of the Phenol-Explorer databank or values measured by HPLC have reported a daily polyphenol intake ranging from 377 ± 15 to [29] 1756.5 ± 695.8 [30] mg/day in many countries [29][30][31]. However, all studies were conducted on healthy subjects, with no study on palliative care cancer patients. Considering that the mean worldwide intake of polyphenols is approximately 1 g/day, the present study detected a habitual daily intake of two to three times less, in agreement with the result reported in a Brazilian population study [29].
In addition, except for sodium, the intake of bers, calcium and of all vitamins analyzed (A, B6, B12, C and E) was below recommended levels. With aging and progression of oncologic disease, modi cations may occur in food consumption due to factors such as loss of appetite, sensory changes in gustatory and olfactory capacity, and social, emotional and economic aspects such as social isolation and depression, with a consequent reduction of the intake and absorption of micronutrients essential for health [32,33].
However, the opposite was observed regarding sodium intake, which was excessive in all groups. This result has been associated with the increased consumption of processed and ultraprocessed foods by the population [34], with 80% of Brazilian older males and 61% of Brazilian older females habitually consuming higher than recommended sodium amounts [34].
Energy and protein consumption was lower than recommended in more than half the patients at the beginning and at the end of the study.
At the beginning and at the end of the study, CG showed lower calorie consumption per Kg than IG1 even with a higher BMI, a higher MAN score and albumin value and better functionality. Despite the di culty in interpreting this nding, we believe that IG1 had a greater consumption per Kg as a form of compensation for its worse basal nutritional status. On the other hand, it has been demonstrated that reduced food intake or low energy intake is independently associated with weight loss in oncologic patients during progression of the disease [35,36].
According to the MAN nutritional screening, most participants had an adequate nutritional status both at the beginning and at the end of the study.
Previous studies have reported higher proportions of malnutrition among cancer patients in palliative care. However, those studies were more heterogeneous regarding the primary location of the tumor, nutritional assessment methods, and functionality [37][38][39]. This divergence may be attributed to the inclusion criteria of the present study.
At the beginning of the study, IG1 subjects had lower screening and nutritional assessment scores determined by the MAN tool and lower BMI and albumin values compared to the other groups. However, at the end of the intervention period, their screening score and MAN results were increased.
Nutritional intervention can reduce the weight loss of patients in an advanced stage of cancer and improve their nutritional status [40].
No differences in body composition were observed here between groups, possibly owing to the short period of intervention. Nevertheless, it should be pointed out that changes in body composition in response to changes in the metabolic demand, physiological changes, aging and alterations due to cancer treatment are frequent among older adults receving palliative care and should be monitored [41].
Laboratory work-up demonstrated progression of oncologic disease. 8-OHdG levels were signi cantly increased in all groups, being possibly associated with the evolution of cancer patients [42].
After the period of intervention, IG1 showed an increase in the levels of the proin ammatory cytokine IL-6 with a concomitant reduction of the antioxidant defense compared to the other groups. These results suggest a worse clinical situation of these patients who already showed greater nutritional impairment at the beginning of the study. Systemic in ammation is associated with worse clinical outcomes, including reduced survival, of cancer patients [43]. GSH and vitamin C play a prominent role in cell protection against cytotoxic and carcinogenic substances [44].
Oxidative stress activates the in ammatory pathways that lead to the transformation of a normal cell into a neoplastic one, also affecting survival, proliferation, invasion, angiogenesis, and resistance to oncologic treatment [45]. Conversely, there is evidence that circulating IL-6 levels may also affect the antioxidant defense system [46]. During the nal phase of the study, IL-6 levels were found to be signi cantly lower in IG2 compared to IG1. In agreement, the levels of MDA, a product of lipid peroxidation, were signi cantly reduced and GSH was increased in the white chocolate group.
We believe that the bene cial action of white chocolate consumption on systemic in ammation and the defense against oxidative stress may be the effect of some not yet studied component. The bene ts of white chocolate intake were also observed in a study by OSTERTAG et al. (2013) [47] conducted on healthy subjects, showing that the consumption of 60 grams of white chocolate in a single intake contributed favorably to platelet activation and to bleeding time compared to bitter chocolate. Since white chocolate does not contain avonoids, the authors suggested that other compounds such as milk serum protein may be responsible for antiplatelet effects [47]. Thus, we may consider white chocolate not to have a placebo effect, except for the evaluation of the polyphenol consumption.
Regarding the QL of the patients, IG1 progressed to higher scores in the functionality domain and subdomains, suggesting that the consumption of chocolate with a higher cocoa content was of bene t in terms of QL.
In a previous study, the authors observed low scores on global and functional health scales, with role functioning showing the worst evaluation, as well as high scores on the symptom scale [48]. In the present srudy, volunteers showed a good QL according to the global health scale and role functiong score, and the initial symptom score was low.
Few studies have analyzed the effect of dark chocolate consumption on QL, but some publications have suggested that supplementation with high cocoa chocolate can be of bene t [49,50].
Strengths and limitations of the study: This was a randomized, controlled study of nutritional intervention with chocolate. To date, we have not found any other studies that evaluated this intervention in older adults with cancer in palliative care. The limitations of the present study were a small number of subjects and a short period of intervention. However, this is an inherent di culty of clinical studies in palliative care. We suggest that further interventions should explore the relations and the underlying causal mechanisms regarding chocolate consumption and its effects on the health and QL of older patients on palliative care.
The present results demonstrate that the consumption of chocolate with a higher cocoa content may contribute to improved nutritional status and functionality among older cancer patients in palliative care with > 70% prognosis of 30-day survival. The consumption of white chocolate was associated with an improvement of oxidative stress parameters.
Good adherence to the consumption of both chocolate types was observed during the study, this being a viable and pleasurable food of easy access contributing to the food supply and well-being of the patients.
Considering that food preferences are highly personal, we believe that nutritional support should also be adapted to the necessities, wishes and preferences of everyone in order to be effective and applicable to the reality of each one. In this respect, nutritional assistance can be an opportunity to aid the patients and their families during treatment.

Declarations
Ethics approval and consent to participate.
The study was approved by the Research Ethics Committee of HC-FMRP-USP (Protocol No. 9614/2015) and all subjects gave written informed consent to participate. This research was registered at www.clinicaltrials.gov (NCT04367493).