Clinical utility of the twiist Automated Insulin Delivery System: first study with twiist

Infusion set occlusions can result in severe hyperglycemia and diabetic ketoacidosis.

“Infusion set occlusions, which impede the delivery of infused insulin, can result in severe hyperglycemia and DKA.21”

• cites [21] Klonoff, 2017, Insulin Pump Occlusions: For Patients Who Have Been Around the (Infusion) Block, J Diabetes Sci Technol sourceSupported

Safety outcomes are comparable to other short-duration AID studies.

“Our results also showed safety outcomes that are comparable to other short-duration AID studies, showing no DKA or severe hypoglycemia events and very little time spent in level 1 and level 2 hypoglycemia.”

• cites [18] sourceSupported
• cites [19] Carlson, 2022, Advanced Hybrid Closed-Loop System Pivotal Trial in Adolescents and Adults with Type 1 Diabetes, Diabetes Technol Ther sourceSupported

Insulin infusion set failures are common among individuals using insulin pump therapy.

“However, insulin infusion set (IIS) failures are common among individuals using insulin pump therapy.3–5”

• cites [3] Wilcox JH , Gray JC , Cooper NR , et al . 696-P: Insulin Pump Infusion Set Failure (IPISF) Management: T1D Exchange Survey. Diabetes 2021;70. doi:10.2337/db21-696-P sourceNot checked — no full text of the source
• cites [4] sourceCould not verify

  The Abstract and Discussion report common skin reactions to CGM and insulin-pump devices, with some patients discontinuing device use. They do not report whether insulin infusion sets fail commonly; searches for infusion-set failure and related terms found no such finding in our copy. (LLM, conf=0.74)

  Experimental — accuracy not yet measured
• cites [5] sourceNot checked — no full text of the source

This finding supports previous studies demonstrating a correlation of percent time in range with more aggressive correction targets.

“This finding supports previous studies demonstrating a correlation of %TIR with more aggressive correction targets.19 20”

• cites [19] Carlson, 2022, Advanced Hybrid Closed-Loop System Pivotal Trial in Adolescents and Adults with Type 1 Diabetes, Diabetes Technol Ther sourceSupported
• cites [20] Tornese, 2023, Safety of Real-life Usage of Advanced Hybrid Closed-Loop System MiniMed 780G in Children With Type 1 Diabetes Younger Than 7 Years Old, Diabetes Care sourcePartly supported

  The results passage (page not provided) reports increased %TIR after Auto Mode began and says that most children later used lower glucose targets. Table 1 tracks both measures over time, but the paper does not demonstrate a correlation between %TIR and more aggressive targets; the initial TIR increase occurred before the reported shift to lower targets. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

The potential Hawthorne effect must be considered a limitation.

“The potential Hawthorne effect must also be considered a limitation.25”

• cites [25] sourceSupported

Current insulin pump systems exhibit substantial variability in occlusion detection times, with delayed alarms-especially at lower basal rates-heightening the risk of unnoticed infusion failures and prolonged hyperglycemic excursions.

“As demonstrated by Freckmann et al, current insulin pump systems exhibit substantial variability in occlusion detection times, with delayed alarms—especially at lower basal rates—heightening the risk of unnoticed infusion failures and prolonged hyperglycemic excursions.23”

• cites [23] Freckmann, 2018, Occlusion Detection Time in Insulin Pumps at Two Different Basal Rates, J Diabetes Sci Technol sourceSupported

Carlson et al reported no diabetic ketoacidosis or severe hypoglycemia events in a cohort of 39 adolescents and 118 adults using the MiniMed 780G system over a 3-month observation period.

“Carlson et al reported no DKA or severe hypoglycemia events in a cohort of 39 adolescents and 118 adults using the MiniMed 780G system over a 3-month observation period.19”

• cites [19] Carlson, 2022, Advanced Hybrid Closed-Loop System Pivotal Trial in Adolescents and Adults with Type 1 Diabetes, Diabetes Technol Ther sourceSupported

The mean difference in the midpoint of the correction range was only 17 mg/dL.

“However, the generalizability of our findings is limited by a small number of participants, and the mean difference in the midpoint of the correction range was only 17 mg/dL.”

After excluding alarms confounded by meals and boluses, 67 occlusion alarms remained for analysis.

“After exclusion of occlusion notifications that may be confounded by meals and insulin boluses, 67 alarms remained for analysis.”

The system demonstrated equal performance in occlusion detection effectiveness across participants with lower total daily insulin delivery.

“Although the system demonstrated equal performance in occlusion detection effectiveness across participants with lower TDD, we cannot generalize this to individuals on much lower TDD (eg, young children) because all participants in our study were adults, and the lowest total daily dose was 26.14 U.”

HbA1c decreased by 0.4±0.5% at 8 weeks in participants with baseline HbA1c >7%.

“Participants with baseline HbA1c >7% (mean 8.1%) showed clinically significant improvements (−0.4±0.5%) at 8 weeks.”

67 occlusion events occurred in 15 participants.

“67 occlusion events occurred in 15 participants, with a median TDD of 58.6 U.”

Mean glucose declined from 143.8±44.7 mg/dL to 141.1±40.6 mg/dL (Δ−2.7±60.4 mg/dL) between 60 and 30 min prior to an occlusion alarm.

“Mean glucose measured at both 60 and 30 min prior to an occlusion alarm declined from 143.8±44.7 mg/dL to 141.1±40.6 mg/dL (Δ−2.7±60.4 mg/dL) (figure 1).”

The study enrolled 20 adults with type 1 diabetes.

“A total of 20 adults with type 1 diabetes were screened and enrolled in the study, with mean HbA1c 6.9%, TDD 52.3 U, and body weight 86.8 kg.”

The percentage of time spent >250 mg/dL did not achieve the CGM target.

“The percentage of time spent >250 mg/dL did not achieve the target.”

No significant change in body weight was observed (Δ−0.09±2.9 kg).

“No significant changes in TDD (Δ1.3±16.2 U) or body weight (Δ−0.09±2.9 kg) were observed.”

At 8 weeks, mean percent time in range (70-180 mg/dL) was 70.3±15.5%.

“At 8 weeks, the mean %TIR was 70.3±15.5%.”

Mean glucose increased from 141.1±40.6 to 149.6±42.5 mg/dL (Δ+8.6±58.8 mg/dL) between 30 min prior to and at the time of an occlusion alarm.

“Mean glucose measured 30 min prior to and at the time of an occlusion alarm increased from 141.1±40.6 to 149.6±42.5 mg/dL (Δ+8.6±58.8 mg/dL).”

Higher maximal basal ratios were associated with greater percent time in range.

“Higher maximal basal ratios were associated with greater %TIR (figure 3).”

144 occlusion alarms were recorded over the study period (0.95 alarms/person-week).

“Across the study period, 144 occlusion alarms were recorded (0.95 alarms/person-week).”

No significant change in total daily insulin delivery was observed (Δ1.3±16.2 U).

“No significant changes in TDD (Δ1.3±16.2 U) or body weight (Δ−0.09±2.9 kg) were observed.”

The percent time in range for participants with a correction range midpoint ≤100 mg/dL was greater than for those with a midpoint >100 mg/dL.

“The %TIR for those with the lower correction range was greater than in those with the higher correction range (figure 4).”

No serious adverse events were reported.

“No serious adverse events were reported.”

The positive correlation between higher maximal basal ratios and increased percent time in range suggests that the algorithm's ability to respond aggressively to increasing glucose levels is associated with improved percent time in range.

“Moreover, the positive correlation between higher maximal basal ratios and increased %TIR suggests that the algorithm’s ability to respond aggressively to increasing glucose levels is associated with improved %TIR.”

The study was not designed to evaluate clinical outcomes directly attributable to occlusion events.

“However, given that the study was not designed to evaluate clinical outcomes directly attributable to occlusion events, these results should be interpreted as supporting feasibility rather than definitive evidence of improved safety or efficacy.”

The single-arm, non-comparative methodology precludes direct attribution of observed improvements to the twiist AID System.

“The single-arm, non-comparative methodology precludes direct attribution of observed improvements to the twiist AID System, and the short study duration does not allow for assessment of long-term glycemic effects, device adherence, or durability of usability outcomes.”

The lack of ethnic, socioeconomic, and geographic diversity in the small, single-site cohort limits generalizability to broader or underserved populations.

“Finally, the lack of ethnic, socioeconomic, and geographic diversity in the small, single-site cohort further limits the generalizability of the findings to broader or underserved populations.”

The small sample size and non-powered design limit the ability to detect statistically significant differences in glycemic outcomes or rare adverse events.

“The small sample size and non-powered design limit the ability to detect statistically significant differences in glycemic outcomes or rare adverse events.”

The generalizability of the findings is limited by a small number of participants.

“However, the generalizability of our findings is limited by a small number of participants, and the mean difference in the midpoint of the correction range was only 17 mg/dL.”

Recruitment from an existing clinic population with higher technology literacy and engagement may bias results toward favorable outcomes.

“Moreover, recruitment from an existing clinic population with higher technology literacy and engagement may bias results toward favorable outcomes, and the structured training and support provided during the study may not reflect real-world initiation, potentially affecting usability and safety outcomes.”

Resolution of occlusions may result in insulin surges, increasing the risk of severe hypoglycemia.

“Conversely, the resolution of occlusions may result in insulin surges, increasing the risk of severe hypoglycemia.22”

• cites [22] sourceNot checked — no full text of the source

The evaluation of occlusion detection under varying insulin requirements provides clinically relevant insight into the system's novel sensing technology and its potential to address a persistent safety concern.

“In addition, the evaluation of occlusion detection under varying insulin requirements provides clinically relevant insight into the system’s novel sensing technology and its potential to address a persistent safety concern associated with insulin pump therapy.”

• AssumptionRests on an assumption: The twiist AID System had not previously been clinically evaluated in adults with type 1 diabetes.
• AssumptionRests on an assumption: The twiist system directly measures insulin volume using acoustic volume-sensing technology.

19 of 20 participants completed all 8 weeks of the study.

“19 participants completed all 8 weeks of the study. One participant withdrew due to time constraints.”

The observed results are in line with expectations for a next-generation AID system.

“These similarities support the conclusion that the observed results are in line with expectations for a next-generation AID system.”

The lack of glycemic data prior to twiist initiation precludes any assessment of glycemic improvement over previous insulin delivery methods.

“Additionally, the lack of glycemic data prior to twiist initiation precludes any assessment of glycemic improvement over previous insulin delivery methods.”

The findings cannot be generalized to individuals on much lower total daily insulin delivery, such as young children, because all participants were adults and the lowest total daily dose was 26.14 U.

“Although the system demonstrated equal performance in occlusion detection effectiveness across participants with lower TDD, we cannot generalize this to individuals on much lower TDD (eg, young children) because all participants in our study were adults, and the lowest total daily dose was 26.14 U.”

The single-center design limits external validity.

“The single-center design limits external validity, as practice patterns and resources may vary across different clinical sites.”

The twiist system can detect occlusions before clinically significant glucose rise regardless of total daily insulin delivery.

“suggests the system’s ability to detect occlusions before clinically significant glucose rise regardless of TDD.”

• AssumptionRests on an assumption: The twiist system directly measures insulin volume using acoustic volume-sensing technology.

Eight-week use of twiist enabled participants to achieve or maintain established glycemic targets.

“Eight-week use of twiist enabled participants to achieve or maintain established glycemic targets and suggests the system’s ability to detect occlusions before clinically significant glucose rise regardless of TDD.”

The twiist system's occlusion detection capability is due to measuring the volume of insulin delivered directly rather than relying on a significant rise in pressure.

“Because the twiist AID System measures the volume of insulin delivered with each dose directly rather than relying on a significant rise in pressure (which takes time), alarms were triggered by measurement of insulin dose within 30 min of a rise in glucose regardless of TDD.”

• AssumptionRests on an assumption: The twiist system directly measures insulin volume using acoustic volume-sensing technology.