This single-arm, nonblinded, single-center pilot study evaluated a low-calorie diet delivered via an eHealth platform and had a study duration of 6 months. The 2 prespecified coprimary outcomes were weight loss of at least 15 kg and diabetes remission, defined as HbA_1c below 6.5% without the use of blood glucose-lowering medication. These coprimary outcomes were selected in accordance with the DiRECT trial, in which weight loss of 15 kg was chosen because previous evidence suggested that this magnitude of weight loss is strongly associated with type 2 diabetes remission. For individuals with a baseline body weight of less than 80 kg, the weight-loss target was 15% of baseline body weight, rather than 15 kg. For participants in whom further weight loss would have resulted in a BMI below 23 kg/m², weight loss was discontinued, and weight maintenance was recommended.

The findings of this study are intended to inform the design of a subsequent randomized controlled trial comparing eHealth-based and face-to-face care.

This pilot study was conducted in Umeå, Sweden, between April and December 2021 and is reported in accordance with the CONSORT (Consolidated Standards of Reporting Trials) 2010 checklist for reporting pilot and feasibility trials (Checklist 1) [12].

Study participants were recruited through advertisements in local newspapers. Additionally, primary care physicians, diabetes nurses, and dietitians were informed about the study at local diabetes meetings and were encouraged to inform their patients about the possibility of participating. Study participants were not involved in the design or conduct of our research.

Eligible participants had been diagnosed with type 2 diabetes within the previous 6 years, were aged 20 to 70 years, had a BMI over 27 kg/m², and had an HbA_1c of 6.5% to 11.9% during the last 12 months without diabetes medication. Exclusion criteria were insulin treatment, weight loss of more than 5 kg in the last 6 months, treatment with weight-loss medications, a diagnosed eating disorder, estimated glomerular filtration rate below 30 mL/min/1.73 m², substance abuse, known cancer, myocardial infarction within the previous 6 months, severe heart failure (New York Heart Association [NYHA] class III), pregnancy or planned pregnancy, and treatment with antipsychotic drugs. Participants were asked to self-report the presence of neuropathy or retinopathy. After study initiation, the upper age limit was revised from 65 to 70 years to enable enrollment of 10 participants. No interim analyses were planned or conducted.

Before study initiation, we decided to include 10 participants in this pilot study.

The study participants replaced all regular food with the TDR formula Modifast (820‐890 kcal/d) for 12 weeks. To counteract constipation, psyllium husk was used as a dietary supplement as needed.

After 12 weeks of TDR, participants entered a structured food reintroduction phase (weeks 12‐18). During this phase, 1 TDR meal at a time was reintroduced.

These were gradually replaced with regular food based on participant preference. The dietary plan included 4 meals per day, with portion guidance provided for breakfast, lunch, dinner, and snacks. The target daily energy intake was 1500 to 1600 kcal for women and 1900 to 2100 kcal for men. Participants received dietary counseling focused on selecting whole grain products, low-fat dairy, foods labeled with the green keyhole symbol, unprocessed meats, and a high intake of vegetables, fruits, and berries. Guidance was also provided on managing potential increases in appetite during this transition. In the subsequent weight maintenance phase (weeks 19‐26), participants followed a fully food-based diet with individualized energy prescriptions from a dietitian to support weight stability. Some participants opted to continue replacing 1 daily meal with a TDR product during this phase.

All participants received a scale (UC-325BLE Continua Body Scale from A&D Medical, Ann Arbor, MI, USA), a finger-stick blood glucose meter (Accu-Chek Instant from Roche, Basel, Switzerland), and a blood pressure monitor (UA-651BLE Continua Arm Blood Pressure Monitor from A&D Medical, Ann Arbor, MI, USA). Participants were asked to measure body weight, blood glucose, and blood pressure every morning. The scale, glucose meter, and blood pressure monitor transmitted their measurements via Bluetooth to the Telia Health Monitoring mobile phone app on the study participant’s mobile phone. From there, the data were sent to the project’s physician, nurse, and dietitian. The results of the measurements were checked by the research nurse every weekday using the application’s secure website for result management. When action was required due to abnormal measurement results, participants received feedback directly or the information was passed on to the doctor or dietitian for follow-up. The participants had a scheduled video visit with the study physician every 2 weeks for the first 3 months. Thereafter, study participants met the study dietitian every 2 weeks until month 4 and then once a month until the end of the study. During the video visits, home measurements of body weight, fasting blood glucose, and blood pressure were discussed.

At the beginning of the food reintroduction phase, participants received an activity tracker (Vivosmart 4 Activity Tracker from Garmin, Olathe, Kansas) that worked together with the Telia Health Monitoring app. During the 12 weeks of TDR, participants were advised not to increase their physical activity. At the beginning of the food reintroduction phase, participants were advised to reach and maintain their sustainable maximum of up to 15,000 steps per day.

HbA_1c was measured monthly using finger-prick blood samples collected by participants at home as part of the intervention. The samples were sent via regular mail to the Department of Clinical Chemistry at Norrland University Hospital for analysis. The HbA_1c results were subsequently reviewed and discussed with study personnel during scheduled video consultations.

All diabetes and blood pressure medications were discontinued on the day the TDR was started because blood glucose levels fall fast, and blood pressure is likely to fall with strict calorie restriction. Medication was reintroduced, as described below, based on blood pressure, blood glucose, and HbA_1c measurements.

The study participants were instructed to avoid salt intake. Blood pressure medication was restarted if systolic blood pressure was over 140 mm Hg on repeated measurements (over 165 mm Hg during the first 2 wk of TDR). To simplify treatment decisions, only systolic blood pressure was used.

Oral hypoglycemic agents were considered only when participants had intolerable hyperglycemic symptoms (polyuria and polydipsia) or a random blood glucose level above 20 mmol/L. If insufficient weight loss led to worsening glycemia, adherence to the low-calorie diet was addressed. If weight loss was adequate but glycemic control remained unsatisfactory, initiation of oral hypoglycemic agents was considered.

If a weight gain of more than 2 kg occurred during the weight loss maintenance phase, diet replacement was recommended for 1 to 2 meals per day for 4 weeks, and a prescription for orlistat was offered. In cases of a weight gain of more than 4 kg or a total weight loss of less than 15 kg from the starting weight, or if diabetes returned, TDR for all meals was recommended for 4 weeks, and thereafter regular meals were reintroduced for 2 to 4 weeks. In addition, orlistat was offered.

All outcome measurements were performed at baseline before initiation of the intervention and at 3 and 6 months at the Clinical Research Center, Umeå University Hospital. Weight was measured on the same scale in light clothing without shoes, and length on a digital height-measuring gauge. Waist circumference was measured midway between the iliac crest and the lower rib. Blood pressure was measured in the right arm with an automatic blood pressure monitor after 5 minutes of rest. Three blood pressure readings were obtained; the first was deleted, and the average pressure of the other 2 was registered.

Fasting venous blood samples were collected and analyzed at the Department of Clinical Chemistry, Umeå University Hospital, using routine clinical laboratory methods.

B-HbA_1c was analyzed using high-performance liquid chromatography with the TOSOH G11 system. Fasting plasma glucose was analyzed by spectrophotometry using the Cobas Pro c503 system and GLUC3 reagent (Roche Diagnostics Scandinavia AB). Glycemic status was classified according to WHO criteria as follows: diabetes, fasting plasma glucose greater than or equal to 126 mg/dL (7.0 mmol/L); impaired fasting glucose (IFG): fasting plasma glucose 110 to 125 mg/dL (6.1‐6.9 mmol/L). Plasma insulin and C-peptide were analyzed by immunometry using the Cobas Pro e801 with Insulin and C-peptide reagents (Roche Diagnostics). The lipid profile, including plasma cholesterol, triglycerides, and high-density lipoprotein cholesterol, was analyzed by spectrophotometry using the Cobas Pro c503 with Cholesterol Gen.2 (CHOL2), Triglycerides (TRIGL), and HDL-Cholesterol Gen.4 (HDLC4) reagents, respectively. Low-density lipoprotein cholesterol was calculated using the Friedewald formula. Liver enzymes, alanine transaminase (ALT) and aspartate aminotransferase (AST), were analyzed by spectrophotometry using the Cobas Pro c503 analyzer with Alanine Aminotransferase acc. to IFCC II (ALTP2) and Aspartate Aminotransferase acc. to IFCC II (ASTP2) reagents, respectively. C-reactive protein (CRP) was analyzed by immunoturbidimetry using the CRP4 reagent (Roche Diagnostics).

An oral glucose tolerance test (OGTT) was performed in the fasting state, during which participants ingested 75 g of glucose dissolved in 200 mL of Topstar solution. Glucose, insulin, and C-peptide levels were measured at 0, 30, 60, 90, and 120 minutes. Glucose tolerance was classified according to the WHO criteria as follows: diabetes, 2-hour plasma glucose greater than or equal to 200 mg/dL (11.1 mmol/L); impaired glucose tolerance (IGT): 2-hour plasma glucose 140-199 mg/dl (7.8‐11.0 mmol/L).

Respiratory quotient (RQ) and resting energy expenditure (REE) were measured using indirect calorimetry (Quark resting metabolic rate, COSMED Nordic ApS). Assessments were performed after an overnight fast, with participants in a supine position. Measurements were obtained 30 minutes before the OGTT, following 20 minutes of rest, and during the final 30 minutes of the OGTT. During each measurement, participants lay under a ventilated hood while the volume of oxygen consumption (VO₂) and the volume of carbon dioxide production (VCO₂) were recorded. The last 10 minutes of each 30-minute recording period were used to estimate RQ and REE, with the requirement that variability during this interval was less than 10% to ensure measurement stability. The RQ was calculated as the ratio of VCO₂ to VO₂, with expected values ranging between 0.7 and 1.0, indicating predominant fat oxidation at lower values (RQ≈0.7) and predominant carbohydrate oxidation at higher values (RQ≈1.0). Metabolic flexibility was assessed by comparing fasting RQ values to RQ values following glucose ingestion, reflecting the body’s ability to shift from fat to carbohydrate oxidation. REE was calculated from fasting measurements using the Weir equation: REE (kcal/d) = 1440 × (3.941 × VO₂ [L/min] + 1.106 × VCO₂ [L/min]). No changes to the measurements were made after the pilot study began.

Data are presented as mean (SD) or median (IQR). The trapezoidal rule was used to calculate the area under the curve (AUC). Data over time from baseline to 6 months were analyzed using the Wilcoxon signed-rank test. A 2-sided P value of <.05 was considered statistically significant. All analyses were performed using SPSS Statistics (version 28.0; IBM Corp).

Ethical approval was granted by the Swedish Ethical Review Authority (Dnr 2021‐00318). The study was prospectively registered at ClinicalTrials.gov (NCT04805996).

This pilot study evaluated whether a TDR intervention supported by eHealth could induce clinically meaningful weight loss and achieve diabetes remission in individuals with type 2 diabetes. Of 10 participants, 60% (n=6) achieved the coprimary outcome of at least 15 kg of weight loss, while the remaining 40% (n=4) lost more than 10 kg. At 6 months, all participants achieved the coprimary outcome of diabetes remission without glucose-lowering medication.

In this study, daily home monitoring of body weight and fasting glucose provided real-time data, enabling the study team to identify insufficient weight loss or weight regain and provide timely support. Daily fasting glucose monitoring also allowed early detection of deteriorating glycemic control, making discontinuation of all glucose-lowering medication a safe component of the intervention.

The exclusive use of eHealth support was associated with high engagement and adherence, consistent with previous studies showing that eHealth interventions improve weight loss and glycemic control. A systematic review found eHealth interventions more effective than control interventions or no care, and comparable to face-to-face interventions for promoting weight loss and weight maintenance [9].

Regarding diabetes remission, defined as HbA_1c below 6.5% without glucose-lowering medication, our findings are consistent with previous studies showing that substantial weight loss is the primary driver of diabetes remission, particularly among individuals with a shorter duration of type 2 diabetes. The Counterpoint study first demonstrated that intensive calorie restriction could restore normoglycemia in individuals with recently diagnosed type 2 diabetes, while the subsequent Counterbalance study highlighted the importance of preserved β-cell function and shorter diabetes duration for achieving remission [13,14]. These findings were later confirmed in larger intervention trials. The DIADEM-I trial, which enrolled adults with type 2 diabetes of 3 years duration or less, reported a remission rate of 61% after 12 months. In comparison, the DiRECT trial included participants with type 2 diabetes of up to 6 years duration and reported remission in 57% (16/28) of those achieving weight loss of 10 to 15 kg and in 86% (31/36) of those losing more than 15 kg at 1 year [6,15]. In our study, all participants achieved diabetes remission after 6 months. Although the small sample size precludes direct comparison, the shorter intervention period, relatively short diabetes duration, and substantial weight loss may explain the higher remission rate.

According to HbA_1c, all participants in our study achieved diabetes remission by the end of the study. However, fasting plasma glucose indicated normoglycemia in only 4 participants, IFG in 5, and diabetes in 1. Similarly, the OGTT showed diabetes in 3 participants and IGT in 7. This discrepancy between HbA_1c and glucose-based measures likely reflects the different aspects of glucose metabolism captured by each marker. HbA_1c reflects average glycemia over the preceding 8 to 12 weeks, fasting glucose primarily reflects hepatic glucose regulation, and 2-hour postload glucose levels provide insight into postprandial glucose metabolism and β-cell function. During early metabolic recovery, as seen in dietary intervention studies, hepatic insulin sensitivity and fasting glucose may normalize before β-cell function is fully restored, leading to persistently elevated 2-hour glucose levels despite normalized HbA_1c and fasting glucose. Previous studies have also shown that HbA_1c alone may underestimate glucose dysregulation, particularly in individuals with IGT [16,17]. Therefore, remission in this study should be interpreted according to consensus HbA_1c criteria rather than as complete normalization of glucose metabolism.

In our study, the greatest reduction in fasting glucose occurred during the first week, as demonstrated by daily home monitoring. Because this preceded substantial weight loss, it is likely to have been driven primarily by caloric restriction. Similar findings were reported by Lim et al [13], who showed that fasting glucose normalized within 7 days of severe caloric restriction (600 kcal/d), accompanied by normalization of hepatic glucose production and a marked reduction in hepatic insulin resistance.

Insulin sensitivity also improved markedly over the 6-month intervention. This contrasts with Lim et al [13], who reported no improvement in insulin sensitivity after 8 weeks of severe caloric restriction, with a mean weight loss of 15 (SD 4) kg among 11 individuals who achieved diabetes remission. However, our group has previously shown that 12 weeks of diet-induced weight loss improves peripheral insulin sensitivity in individuals with type 2 diabetes [18].

Although insulin sensitivity improved in our study, insulin secretion did not improve. This contrasts with Lim et al [13], who reported that diabetes remission was primarily driven by a rapid improvement in insulin secretion, evident within the first week and nearly normalized by week 8. A possible explanation is the different methods used to assess β-cell function. Lim et al [13] used a stepped insulin secretion test with arginine, a more direct and sensitive measure of β-cell function, whereas we used the insulinogenic index derived from OGTT glucose and insulin measurements, which may be less sensitive to subtle or early changes in β-cell responsiveness. Therefore, it remains unclear whether diabetes remission requires improvements in both insulin secretion and insulin sensitivity or whether improved insulin secretion alone is sufficient.

REE decreased progressively over the 6-month intervention, with a median reduction of 387 (IQR 381-536) kcal/day, accompanied by a mean weight loss of 16.9 (SD 5.4) kg. This exceeds the reductions reported in the POUNDS LOST study, where REE decreased by approximately 100 kcal/day in men and 55 kcal/day in women after mean weight losses of 7.6 (SE 0.34) kg and 5.89 (SE 0.27) kg, respectively, at 6 months [19].

The greater reduction in REE observed in our study may partly reflect the fact that postintervention assessments were performed immediately following the weight-loss phase, when some participants were likely still in negative energy balance. Consistent with this interpretation, Martins et al [20] showed that the apparent metabolic adaptation observed immediately after weight loss was substantially attenuated after weight stabilization, suggesting that part of the reduction in REE reflects ongoing negative energy balance rather than a persistent suppression of resting metabolism. Therefore, the reduction in REE observed in our study should be interpreted with caution, as it may partly represent a transient physiological response to active weight loss rather than a long-term change.

The RQ is the ratio of carbon dioxide produced to oxygen consumed during metabolism and provides insight into the balance between carbohydrate and fat oxidation [21,22]. An RQ of 1.0 indicates exclusive carbohydrate oxidation, whereas an RQ of 0.7 indicates exclusive fat oxidation, with intermediate values reflecting mixed substrate utilization. Metabolic flexibility refers to the ability to adapt fuel oxidation to fuel availability by switching between carbohydrate and fat oxidation [22,23]. Impaired metabolic flexibility is common in obesity and type 2 diabetes [24,25]. At baseline, the low fasting RQ (0.67) indicated predominant fat oxidation, while the modest increase after glucose ingestion (0.74) suggested impaired metabolic flexibility. After 6 months, fasting RQ increased to 0.78, indicating greater carbohydrate oxidation in the fasting state, contrary to our hypothesis. However, postglucose RQ increased more markedly to 0.86, consistent with improved metabolic flexibility. These findings are consistent with previous studies showing that improvements in insulin sensitivity following lifestyle intervention are accompanied by greater increases in RQ during glucose or meal stimulation rather than consistent changes in fasting RQ [26]. Together, our findings suggest that weight loss improved the capacity to oxidize carbohydrates when they were available. We did not observe increased fasting fat oxidation, possibly because fat oxidation was already high at baseline. In addition, some participants may still have been in negative energy balance at the postintervention assessment, which could have influenced the RQ measurements [27].

A key strength of this pilot study was the high level of participant engagement and adherence, supported by daily self-monitoring and continuous eHealth feedback from the study team. This real-time interaction likely contributed to sustained motivation and compliance. The diet intervention, including TDR, was evidence-based and designed to minimize barriers to adherence by reducing the need for food preparation, while clearly defined treatment goals promoted consistency.

Several limitations should be acknowledged. This pilot study included only 10 participants, all with relatively short-duration type 2 diabetes (mean 2.8, SD 1.3 y), which is known to increase the likelihood of achieving remission. The wide age range may have increased variability. Although P values are reported, the small sample size may have led to an overestimation of statistical significance. The intervention and follow-up were relatively short. Recruitment through local newspaper advertisements may have introduced selection bias by attracting highly motivated individuals, limiting generalizability to routine clinical populations. In addition, only 1 participant was a woman and none were from non-White ethnic backgrounds, limiting the generalizability of the findings. Women may face additional barriers to participating in intensive lifestyle interventions, including caregiving responsibilities, family commitments, time constraints, and logistical challenges [28,29]. Future randomized studies should therefore use targeted recruitment strategies, including recruitment through primary care and community outreach, flexible scheduling, and prespecified recruitment targets, to improve the representation of women and ethnically diverse populations and enhance external validity [28,30,31].

Glucagon-like peptide-1 receptor agonists and tirzepatide were not used, as pharmacologically induced weight loss would have precluded evaluation of medication-free diabetes remission. As incretin-based therapies become more widely used, digitally delivered dietary interventions may primarily appeal to individuals seeking medication-free remission or those with limited access to, or contraindications to, these therapies.

Given the increasing prevalence of obesity and type 2 diabetes, public health efforts should prioritize interventions that are both accessible and scalable. If eHealth-based interventions prove effective in larger and more diverse populations, they could provide a sustainable and cost-effective complement or alternative to in-person care while improving access for people living in rural or underserved communities.

This pilot study demonstrates that type 2 diabetes of short duration may enter remission through substantial weight loss, as reflected by HbA_1c levels falling below 6.5% without the use of diabetes medication. However, long-term remission appears to depend on sustained weight loss, as relapse is likely with weight regain. While all participants met the HbA_1c criterion for remission, postload glucose levels from the OGTT remained in the impaired or diabetic range in several individuals, highlighting a discrepancy between HbA_1c and dynamic glucose regulation. This suggests that HbA_1c alone may underestimate residual glycemic impairment and that full metabolic recovery may not yet be complete. These findings underscore the potential of structured total replacement diet interventions, supported by eHealth tools, to achieve clinically meaningful remission in type 2 diabetes, even outside conventional health care settings. However, further research is needed to confirm these findings in larger and more diverse populations over a longer period.