Cohen 2018.

Study characteristics
Methods	Design: parallel design, 2‐arm, open‐label randomised controlled trial Setting: single‐centre study at the University of Alabama at Birmingham, USA. Women with ovarian or endometrial cancer, recruited between October 2015 and April 2017, from the University of Alabama Gynecologic Oncology clinic and from other treatment centres via physician referral, flyers, local television advertisements and news articles. Follow‐up: 12 weeks Fasting serum concentrations of glucose, insulin, C‐peptide, insulin‐like growth factor‐I, insulin‐like growth factor binding protein‐1, and β‐hydroxybutyrate, and body composition obtained at baseline and 12 weeks.
Participants	Number of participants enroled 73 women randomised; 57 of whom went onto their randomised diet and only 45 of these completed trial. 18/57 who completed baseline testing had endometrial cancer and BMI ≥ 25. 13 kg/m2 of these participants completed the trial, with 9 randomised into the ketogenic diet intervention arm, and 4 into ACS diet control arm. Inclusion criteria Women aged ≥ 19 years History of endometrial or ovarian cancer Measurable disease or elevated CA‐125 BMI > 18.5 kg/m2 English speaking or reading Able to sign consent and willing to be randomised and adhere to the assigned protocol Exclusion criteria Pre‐existing medical conditions: uncontrolled hypertension, unstable angina or myocardial infarction or cerebrovascular accident in 6 months prior to study, congestive heart failure, serious infectious diseases, chronic hepatitis, cirrhosis, chronic malabsorption syndrome, chronic pancreatitis, chronic lung disease, major depressive or psychiatric disorder. Current or medical condition that affects bodyweight such as uncontrolled hypo‐ or hyperthyroidism. Taking any of the following medications: antipsychotic agents, monoamine oxidase inhibitors, antibiotics for HIV or tuberculosis, weight loss medications or have taken weight loss medications in last 6 months. Currently dieting Baseline participant characteristics These characteristic refer solely to the 13 participants with endometrial cancer and BMI ≥ 25 kg/m² who completed the trial. Mean age 59.9 (SD 11.1) years, mean BMI 37.4 (SD 9.9) kg/m2. Ethnic background of this specific group of study participants was not provided. Comorbidities included hypertension (46.2%), hyperlipidaemia (38.5%), hyperglycaemia (7.7%), hypothyroidism (30.8%) and sleep apnoea (15.4%). Histological type of endometrial cancer was not provided, stage of cancer was: Ia (61.5%), Ib (23.1%), II (7.7%) and IIIb (7.7%). The ECOG PS was not provided, all had undergone primary treatment (details not provided) and 2 were receiving adjuvant treatment during trial (details not provided).
Interventions	Ketogenic diet intervention arm Ketogenic diet (low‐carbohydrate and high‐fat) with advice provided by dietitian, weekly counselling via telephone or email, sample meal plans and recipes were provided. Urinary ketones measured daily for 2 weeks and weekly thereafter. Quote: "The KD [ketogenic diet]had a macronutrient distribution of ∼5% of energy from carbohydrate (≤20 g/d), 25% energy from protein (≤100 g/d), and 70% energy from fat (≥125 g/d). Carbohydrate foods were limited to nonstarchy vegetables such as salad greens, broccoli, and summer squash. Permitted protein foods included meat, poultry, eggs, and fish, provided that they were neither breaded nor battered. Fat‐containing foods included olive and coconut oils, avocados, butter, olives, cheese, cream, and small amounts of nuts. KD participants were instructed to avoid all grains and grain products, starchy vegetables, and fruit. Total energy intake was not restricted for either the ACS or KD. A registered dietitian provided diet‐specific nutrition education to each participant immediately after baseline testing. Additional counseling was provided via phone and e‐mail on a weekly basis, and included distribution of sample meal plans and recipes. When necessary, based on feedback from the participant, diet recommendations were further individualized to enhance adherence, assessed via weekly food record reviews. In addition, KD participants were also asked to measure urinary ketones with Ketostix (Bayer AG, Leverkusen, Germany). They submitted smartphone photographs of their strips via e‐mail daily for the first 2 wk, and thereafter on a weekly basis." ACS control arm Low‐fat diet with advice provided by dietitian, weekly counselling via telephone or email, sample meal plans and recipes were provided. Quote: "The ACS diet consisted of general guidelines to encourage intake of antioxidants and fiber, while reducing consumption of saturated fat and added sugars … Total energy intake was not restricted for either the ACS or KD. A registered dietitian provided diet‐specific nutrition education to each participant immediately after baseline testing. Additional counseling was provided via phone and e‐mail on a weekly basis, and included distribution of sample meal plans and recipes. When necessary, based on feedback from the participant, diet recommendations were further individualized to enhance adherence, assessed via weekly food record reviews."
Outcomes	Primary outcomes Overall survival: 1 of the 18 participants with endometrial cancer with BMI ≥ 25 kg/m² died during the study period (given as reason for withdrawal). It was not reported which group the participant was allocated to, or whether the death was related to cancer or not. Adverse events: not reported. Secondary outcomes Recurrence‐free survival: not reported. Cancer‐specific survival: not reported.  Weight loss: change in weight from baseline to 12 weeks reported.  Cardiovascular and metabolic event frequency: not reported. QoL: change in QoL from baseline to 12 weeks reported using SF‐12 Physical Health Component change score and Mental Health Component change score (responses from 5/9 participants in ketogenic diet group and 3/4 participants in ACS group). Power Comment: power calculation performed, and there were sufficient details to allow it to be replicated. Quote: "Statistical power analysis indicated that 25 individuals per group were required to have 80% power to detect a 2.7 ± 4.6 μU/mL difference in fasting insulin, through the use of a 2‐sided paired t test and a significance alpha level of 0.05."
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Comment: computer‐generated blocked randomisation scheme created prior to study opening. Quote: "Participants were randomly assigned, through the use of a computer‐generated blocked randomization scheme."
Allocation concealment (selection bias)	High risk	Comment: group allocation performed by the project co‐ordinator. Quote: "The project co‐ordinator enrolled and assigned participants to their intervention."
Blinding of participants and personnel (performance bias) All outcomes	High risk	Comment: participants and personnel were unblinded. Quote: "Because this was a diet intervention study, it was not possible for participants or study personnel to be blinded to group assignment."
Blinding of outcome assessment (detection bias) All outcomes	High risk	Comment: outcome assessments performed by a mix of blinded and unblinded personnel. Quote: "Research team members were provided only with subject IDs and demographics as needed to complete measurements/analyses." Quote: "Study coordinator, physician, and dietitian were aware of assignment, but DXA [dual‐energy X‐ray absorptiometry] technicians, nursing staff, and laboratory staff were blinded."
Incomplete outcome data (attrition bias) All outcomes	High risk	Follow‐up Entered into the study: 73 participants; 36 in ACS control arm and 37 in ketogenic diet intervention arm (unknown number of these were women with endometrial cancer with BMI ≥ 25 kg/m2). 57 participants completed baseline testing; 18 women with endometrial cancer with BMI ≥ 25 kg/m2. Withdrew from the study: 5 (no information given for which arm) Completed the study: 4 in ACS control arm, 9 in ketogenic diet intervention arm. Missing data: for QoL, analysis performed for 3 participants in ACS control arm and 5 participants in ketogenic diet intervention arm. Intention‐to‐treat analysis Comment: not performed.
Selective reporting (reporting bias)	Low risk	Comment: protocol not published, but trial registered prospectively on ClinicalTrials.gov and all prespecified outcomes reported.
Other bias	Low risk	Source of funding: (quote) "Supported by American Institute for Cancer Research, UAB Comprehensive Cancer Center, Nutrition Obesity Research Center grant P30DK56336, and Diabetes Research Center grant P60DK079626." Ethical approval: (quote) "UAB’s Institutional Review Board approved the study" Conflicts of interest: no conflicts of interest reported.