Major weight loss and taste changes are well documented in patients with hematological cancer during chemotherapy. It has previously been documented, that such patients have preferences for much umami, a little sweet, sour and salt, and no bitter. The purpose of the study was to convert these results into real diets. Patients participated in two sensory pilot studies (n=10), where dishes were tested for preferences before and after chemotherapy. From these results four dishes were selected and tested on 32 patients in 30 days in a cross-over design. The diets resulted in a beneficial and statistical significant difference in weight development (p= 0.0008), with 1.2 ± 1.9 kg (+2%) in the intervention period and -2.8 ± 5.2 kg (-4%) in the control period. This difference persisted after sensitivity analysis (± 10%) p= 0.005. However, the nutritional intake was still low in both periods, and the treatment with cytarabin turned out to be a major confounder as dosage was significantly higher in the control period.

The study was approved by the regional scientific ethical committee, and was registered in ClinicalTrial.gov NTC no………...

Patients were selected consecutively with acute and chronic leukemia and lymphoma, Hodg-kin's lymphoma, and Non-Hodgkin's lymphoma in treatment with chemotherapy.

Eligible patients were aged >18 years, and only patients who were able to com-municate and understand written and oral information and receive food orally were included. The special diets were developed at the Central Kitchen at Rigshospitalet Initially, two single blinded pilot studies of five patients were performed, with the same inclusion and exclusion criteria as the main study. In the first study patients tested 17 dishes with a taste-mix of dominating umami, a little sweet, sour and salt, and no bitter. In the second round six dishes were selected based on the results from the first round. All dishes had a high energy-density, with a protein content of at least 5 g/100 g. Every dish was tested both before and after chemotherapy. Patients rated acceptability on 9-point scales.

The clinical controlled cross-over design with two periods of four weeks, intervention and control, tested four dishes selected after the second pilot study. 32 patients were included in that trial, and they started with their habitual diet as control and then switched to recommendation of introduction of the four dishes selected.

The patients registered their weight every time they were at the hospital. During the intervention period the patients received a recipe booklet with the four dishes and a noted every time they ingested one of these. Compliance measurements were based on this registration. Patients had access to the dishes when in-patients, but they did their own cooking at home. It was possible to buy the similar dishes as pre-prepared meals in the supermarkets. The dishes in the final study were: Chili con Carne, Chicken in Curry, Curry Soup and Pasta Carbonara with added curry.

Data were analyzed with R (version 2.15.3). In histograms used to show the degree of acceptability for the dishes in the pilot studies, answers above 5 were graded as high grade of acceptance. To present descriptive statistics, the results were expressed as mean ± standard deviation (SD), median (interquartile range (IQR)) and range. As the data did not show a normal distribution, Wilcoxon rang-sum tests (paired and unpaired) were used. Analyses were carried out both according to intention-to- treat (ITT) and per-protocol (PP) principles. As many of the data sampled, especially the subjective ones, sensitivity analysis (± 10% correction of the results) were performed.

Inclusion Criteria:

• Hematological cancer
• Ongoing chemotherapy
• Ability to eat by mouth
• Informed consent

Exclusion Criteria:

• Lack of the ability to understand the informations given
• Lack of the ability to understand the conditions of the experiment