Endocrinology · Medical Literature
The CRUMB study: closed-loop response to unannounced mixed and carbohydrates-rich breakfasts. A randomized controlled crossover pilot-study in a cohort of adolescents with type 1 diabetes.
Abstract
Unannounced meals pose a major challenge to type 1 diabetes patients. This study compared the performance of two automated insulin delivery (AID) algorithms, the Medtronic MiniMed 780G's proportional-integral-derivative (PID) and the Tandem t: slim X2's model predictive control (MPC), following unannounced breakfasts (ClinicalTrials.gov identifier: NCT07455643, retrospectively registered the 3rd of February 2026). In a randomized crossover study, we enrolled 20 children between 11 and 18 years using the MiniMed 780G or the Tandem t: slim X2 AID system. Endpoints included 2-hour and 4-hour blood glucose difference (ΔBG), glucose peak, time to peak, and time spent above (TAR), below (TBS), in range (TIR), and in tight range (TITR). Two meals were tested: a carbohydrate (CHO) meal and a mixed one, both containing 30 g of carbohydrates, and additional 15 g proteins for the mixed meal. Announced (AM) and unannounced (UM) meals were analyzed. AM, compared to UM, showed significant higher 2-hourΔBG; conversely 4-hourΔBG did not differ significantly. In CHO AM, algorithms were comparable. In mixed AM, PID gained lower peaks and TAR, with higher TIR and TITR. In UM, PID obtained lower 2-hourΔBG than MPC, with reduced peak, TAR, and improved TIR and TITR. Differences in 4-hourΔBG and time-at-peak were not significant. Both AID algorithms mitigated postprandial glycemia and returned glucose to baseline levels within 4 h without safety concerns. The PID demonstrated higher reactivity to unannounced meals, while performances were comparable when meal boluses were properly announced. NCT07455643.
Clinical relevance
In adolescents with type 1 diabetes using commercially available automated insulin delivery (AID) systems, does the Medtronic MiniMed 780G's proportional-integral-derivative (PID) algorithm perform differently from the Tandem t:slim X2's model predictive control (MPC) algorithm in managing postprandial glycemia following unannounced carbohydrate-only and mixed (carbohydrate plus protein) breakfasts? __SNAPSHOT__ {"studyDesign":"Randomized controlled crossover pilot study (ClinicalTrials.gov: NCT07455643; retrospectively registered February 3, 2026)","setting":"Not reported (institution type and country not specified in the abstract)","sampleSize":"n=20 adolescents; algorithm allocation not further subdivided in the abstract (pilot study)","population":"Children and adolescents aged 11–18 years with type 1 diabetes who were existing users of either the Medtronic MiniMed 780G or Tandem t:slim X2 AID system","intervention":"Two standardized breakfast conditions tested on each AID system: (1) a carbohydrate-only (CHO) meal containing 30 g carbohydrate; and (2) a mixed meal containing 30 g carbohydrate plus 15 g protein. Both meal types were administered as announced (AM) and unannounced (UM) variants.","comparator":"Cross-device comparison: Medtronic MiniMed 780G (PID algorithm) vs. Tandem t:slim X2 (MPC algorithm); within-subject crossover design also compared announced vs. unannounced conditions for each meal type","primaryOutcome":"2-hour blood glucose difference (ΔBG) and 4-hour ΔBG (pre- vs. post-meal); secondary endpoints included glucose peak, time to peak, time above range (TAR), time below range (TBR), time in range (TIR; 70–180 mg/dL), and time in tight range (TITR; 70–140 mg/dL)","followUp":"Acute postprandial observation windows of 2 hours and 4 hours per meal challenge; total study follow-up duration not reported in the abstract","fundingSource":"Not reported"}
Key findings
Announced meals (AM) produced significantly higher 2-hour ΔBG compared with unannounced meals (UM) across both device groups; however, 4-hour ΔBG did not differ significantly between AM and UM conditions—specific p-values, confidence intervals, and absolute ΔBG magnitudes are not reported in the abstract. For the CHO announced meal, the two AID algorithms (PID vs. MPC) demonstrated comparable glycemic performance across all endpoints. For the mixed announced meal, the PID algorithm (MiniMed 780G) achieved lower glucose peaks and lower TAR, with higher TIR and TITR compared with the MPC algorithm (Tandem t:slim X2); exact between-group differences, p-values, and confidence intervals for these comparisons are not reported in the abstract. Findings for unannounced CHO and unannounced mixed meals are referenced in the abstract as partially presented but results are truncated; complete data for these conditions are not available from the abstract alone. No data on hypoglycemia rates, TBR values, adverse events, or NNT/NNH are reported in the abstract. No independently calculated values are presented.
Practice implications
This pilot study provides preliminary signal that, under announced conditions, the MiniMed 780G PID algorithm may offer superior postprandial glucose control (lower peaks, lower TAR, higher TIR and TITR) over the Tandem t:slim X2 MPC algorithm specifically for mixed macronutrient meals containing protein, while the two systems appear equivalent for pure carbohydrate meals. The observation that announced meals produced a significantly higher 2-hour ΔBG than unannounced meals is counterintuitive and warrants careful interpretation: it may reflect pre-bolus timing effects, meal composition interactions with algorithm response, or statistical artifact in a small sample—the abstract does not elaborate on this finding. The equivalence of 4-hour ΔBG between announced and unannounced meals suggests that both systems recover to comparable glycemia by 4 hours regardless of announcement status, which has potential relevance for counseling adolescents who frequently omit boluses. Critical limitations of interpretation include: (1) only an abstract is available—full results, statistical tables, and sub-group data cannot be assessed; (2) the pilot sample (n=20) is underpowered for definitive algorithm comparisons; (3) retrospective registration introduces risk of outcome reporting bias; (4) the abstract is truncated and does not present UM data for either meal type in full. These findings are not sufficient to guide device selection in clinical practice and do not apply beyond the adolescent age range tested. Patients not represented include young children, adults, those with severe hypoglycemia unawareness, those with gastroparesis, and those with HbA1c outside the unstated trial range. __LIMITATIONS__ Small pilot sample (n=20) with insufficient statistical power for reliable between-algorithm comparisons; crossover design mitigates but does not eliminate this concern.,Retrospective trial registration (registered February 2026 for a study published July 2026) raises concern for selective outcome reporting and reduces confidence in pre-specified endpoint integrity.,Abstract is truncated; results for unannounced meal conditions and full statistical outputs (p-values, confidence intervals, effect sizes) are not available for independent evaluation.,Only an abstract was available for this summary; full-text data, methods, randomization procedures, and complete results tables could not be reviewed.,Acute postprandial design (2- and 4-hour windows) with standardized controlled meals in a research setting limits generalizability to free-living conditions where meal composition, portion size, and timing vary substantially.,Patients were pre-existing users of their respective AID systems, introducing potential selection bias and differential device familiarity that could influence glycemic outcomes.,Counterintuitive finding (AM > UM for 2-hour ΔBG) is not explained in the abstract, raising concerns about confounding by pre-meal glucose levels, bolus timing protocols, or carryover effects in the crossover design.,Funding source is not disclosed, precluding assessment of industry-related bias.,Study restricted to ages 11–18; findings cannot be extrapolated to younger children or adults.,No patient-reported outcomes, quality-of-life measures, or long-term glycemic metrics (HbA1c, sensor wear time) are reported. __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ This small randomized crossover pilot (n=20) suggests that both the MiniMed 780G (PID) and Tandem t:slim X2 (MPC) AID systems handle unannounced 30 g carbohydrate breakfasts with comparable 4-hour glucose recovery, but the 780G PID algorithm may offer better postprandial metrics for announced mixed (carbohydrate + protein) meals. Given the pilot scale, truncated abstract, retrospective trial registration, and absence of full statistical data, these findings are preliminary and hypothesis-generating only—device selection for individual adolescents with type 1 diabetes should continue to be guided by existing comparative effectiveness evidence, patient preference, and access considerations until confirmatory data are available.