Diabetes Ther (2018) 9:1269–1277 https://doi.org/10.1007/s13300-018-0436-y ORIGINAL RESEARCH Artiﬁcial Pancreas as an Effective and Safe Alternative in Patients with Type 1 Diabetes Mellitus: A Systematic Review and Meta-Analysis . . . . Xia Dai Zu-chun Luo Lu Zhai Wen-piao Zhao Feng Huang Received: April 13, 2018 / Published online: May 9, 2018 The Author(s) 2018 Methods: Electronic databases were carefully ABSTRACT searched for English publications comparing artiﬁcial pancreas with its control group. Over- Introduction: Insulin injection is the main all daytime and nighttime glucose parameters treatment in patients with type 1 diabetes mel- were considered as the endpoints. Data were litus (T1DM). Even though continuous glucose evaluated by means of weighted mean differ- monitoring has signiﬁcantly improved the ences (WMDs) and 95% conﬁdence intervals conditions of these patients, limitations still (CIs) generated by RevMan 5.3 software. exist. To further enhance glucose control in Results: A total number of 354 patients were patients with T1DM, an artiﬁcial pancreas has included. Artiﬁcial pancreas signiﬁcantly main- been developed. We aimed to systematically tained a better mean concentration of glucose compare artiﬁcial pancreas with its control (WMD - 1.03, 95% CI - 1.32 to - 0.75; P = group during a 24-h basis in patients with 0.00001). Time spent in the hypoglycemic phase T1DM. was also signiﬁcantly lower (WMD - 1.23, 95% CI - 1.56 to - 0.91; P = 0.00001). Daily insulin Enhanced digital features To view enhanced digital requirement also signiﬁcantly favored artiﬁcial features for this article go to https://doi.org/10.6084/ m9.ﬁgshare.6200927. pancreas (WMD - 3.43, 95% CI - 4.27 to - 2.59; P = 0.00001). Time spent outside the euglycemic phase and hyperglycemia phase (glucose [ 10.0 X. Dai L. Zhai W. Zhao Department of Endocrinology, The First Afﬁliated mmol/L) also signiﬁcantly favored artiﬁcial pan- Hospital of Guangxi Medical University, Nanning creas. Also, the numbers of hypoglycemic events 530021, Guangxi, People’s Republic of China were not signiﬁcantly different. Z. Luo Conclusion: Artiﬁcial pancreas might be con- Department of Internal Medicine Education, The sidered an effective and safe alternative to be First Afﬁliated Hospital of Guangxi Medical used during a 24-h basis in patients with T1DM. University, Nanning 530021, Guangxi, People’s Republic of China Keywords: Artiﬁcial pancreas; Glucose control; F. Huang (&) Type 1 diabetes mellitus Institute of Cardiovascular Diseases and Guangxi Key Laboratory Base of Precision Medicine in Cardio-cerebrovascular Diseases Control and Abbreviations Prevention, The First Afﬁliated Hospital of Guangxi T1DM Type 1 diabetes mellitus Medical University, Nanning, Guangxi 530021, WMD Weight mean difference People’s Republic of China e-mail: email@example.com CIs Conﬁdence intervals 1270 Diabetes Ther (2018) 9:1269–1277 Search Strategies INTRODUCTION During this process, we searched for the terms Type 1 diabetes mellitus (T1DM) is still a major ‘‘artiﬁcial pancreas’’, ‘‘bionic pancreas’’, ‘‘closed concern in this new era. This chronic disorder loop glucose control’’, ‘‘type 1 diabetes melli- occurs when beta cells of the pancreas are tus’’, ‘‘diabetes mellitus’’, ‘‘glucose control’’, and destroyed by autoimmune antibodies at a ‘‘glucose monitoring’’, one at a time and in younger age (childhood) or it can even develop combination. Publications comparing artiﬁcial in adults, during their late 30s or 40s . Insulin pancreas with its control group were considered injection has been the main treatment in these relevant to this analysis. patients with T1DM . However, despite con- certed efforts of patients and physicians/care- givers, control of blood glucose has often been Inclusion Criteria difﬁcult to achieve . With the development of new techniques Studies were included if they compared artiﬁcial and devices in clinical medicine, continuous pancreas with its control group (any relevant glucose monitoring has signiﬁcantly improved control group); they reported glucose control conditions of patients with T1DM . To fur- parameters or other outcomes related to glucose ther enhance glucose control in these patients, monitoring (during daytime and overnight/24- an artiﬁcial pancreas has recently been devel- h basis); they involved patients with T1DM. oped . Even though this artiﬁcial pancreas has Exclusion Criteria already been approved for use by the US Food and Drug Administration (FDA) , the exis- Studies were excluded if they did not compare tence and beneﬁts of this ‘‘expected to be’’ artiﬁcial pancreas with its control group; they effective device are not well known to the gen- did not report glucose monitoring parameters as eral population. their endpoints or they reported only daytime In this analysis, we aimed to systematically or overnight measurement, but not both in compare artiﬁcial pancreas with its control combination; they involved patients with type group in terms of effectiveness and safety, dur- 2 diabetes mellitus instead of patients with ing a 24-h treatment of patients with T1DM. T1DM; they were duplicates. METHODS Type of Participants, Endpoints, and Follow-Up Periods Searched Databases This analysis included patients with T1DM. The Medline database of medical publications, The endpoints (during daytime and night- the Cochrane library of randomized controlled time) which were analyzed included: trials, and EMBASE database were carefully (a) Mean and median glucose concentration. searched by the ﬁve authors for English publi- (b) Time spent outside the euglycemic phase cations comparing artiﬁcial pancreas with its (outside the glucose range 3.90 to control group (any control group). Reference 8.0 mmol/L). lists of selective publications (most relevant (c) Time spent in the hypoglycemia phase ones) were also carefully reviewed for appro- (blood glucose \ 3.9 mmol/L). priate articles. In addition, ofﬁcial websites of (d) Insulin required/delivered per day. speciﬁc journals such as Lancet Diabetes and (e) Time spent in the hyperglycemia phase Endocrinology, Diabetes Care, and Cardiovascular (blood glucose [ 10.0 mmol/L). Diabetology were also searched for any relevant (f) Number of hypoglycemic events. publication. Diabetes Ther (2018) 9:1269–1277 1271 The follow-up period ranged from less than RESULTS 1 week to 2 months. Search Outcomes, Main and Baseline Data Extraction, Review, and Statistical Features of the Studies Analysis Eight studies [9–16] were selected to be used in After the search process, which was conducted this analysis as shown in Fig. 1. in accordance with the PRISMA guideline , Table 1 summarizes the main features of the the same reviewers assessed the titles and studies included in this analysis. As per the cri- abstracts and independently selected the most teria of this analysis which required an experi- suitable articles which satisﬁed the inclusion mental and a control group, a total of 354 and exclusion criteria of this analysis and then patients were included (177 patients in each data were extracted. The studies which were group) as shown in Table 1. included in this analysis were judged as having The baseline features are reported in Table 2. low to moderate risk of bias . This is a meta-analysis and therefore incon- Analysis Results sistency across the studies was evident . Hence, heterogeneity was assessed by two sta- Results of this current analysis showed artiﬁcial tistical methods: pancreas to signiﬁcantly maintain a better (a) The Q statistic test, whereby a P value less mean concentration of glucose with WMD or equal to 0.05 was considered statistically - 1.03, 95% CI - 1.32 to - 0.75; P = 0.00001, signiﬁcant. I = 46% compared to the control group during (b) The I statistic test; a high percentage value a 24-h basis. The median glucose concentration indicated high heterogeneity (whereby a was similar in both groups with WMD - 0.30, random effects model was used) and low 95% CI - 1.03 to 0.44; P = 0.43, I =0% as percentage value denoted low heterogene- shown in Fig. 2. ity (whereby a ﬁxed effects model was In addition, the time spent in the hypo- used). glycemic phase (glucose \ 3.9 mmol/L) also Since continuous data was used in this analysis, signiﬁcantly favored artiﬁcial pancreas with i.e., mean and standard deviation (SD), data were evaluated by means of weighted mean differences (WMDs) and 95% conﬁdence inter- vals (CIs). In case the SD value was not pro- vided, but a p value was given, SD was calculated using the formula SD = Hn 9 p 9 (1 – p). The analysis was carried out by RevMan 5.3 software. Publication bias was assessed by visually observing funnel plots. Compliance with Ethics Guidelines This meta-analysis is based on previously con- ducted studies and does not contain any studies with human participants or animals performed by any of the authors. Fig. 1 Flow diagram representing the study selection 1272 Diabetes Ther (2018) 9:1269–1277 Table 1 Main features of the studies included Studies Type of study Patients with AP Patients in control group Total patients (n) (n) (n) Blauw  Randomized crossover 5 5 10 Kropff  Randomized crossover trial 32 32 64 Renard  Single-arm non-randomized 20 20 40 extension Thabit  Randomized crossover 24 24 48 Kovatchev  Randomized crossover 18 18 36 El-Khatib  Randomized crossover trial 39 39 78 Russell  Randomized crossover 20 20 40 Russell  Randomized crossover trial 19 19 38 Total patients 177 177 354 (n) AP artiﬁcial pancreas Table 2 Baseline features of the patients in the studies included Studies Mean age (years) Male (%) HBA1c (%) BMI (kg/m ) DM duration (years) AP/C AP/C AP/C AP/C AP/C Blauw – – – – – Kropff  47.0/47.0 56.0/56.0 8.2/8.2 25.1/25.1 28.6/28.6 Renard  46.3/46.3 45.0/45.0 8.2/8.2 24.9/24.9 28.9/28.9 Thabit  43.0/43.0 54.2/54.2 8.1/8.1 26.0/26.0 29.0/29.0 Kovatchev – – – – – El-Khatib  33.3/33.3 46.0/46.0 7.7/7.7 25.9/25.9 16.9/16.9 Russell  40.0/40.0 40.0/40.0 7.1/7.1 25.0/25.0 24.0/24.0 Russell  9.80/9.80 32.0/32.0 7.8/7.8 17.8/17.8 5.00/5.00 AP artiﬁcial pancreas, C control group, HBA1c glycosylated hemoglobin, BMI body mass index, DM diabetes mellitus WMD - 1.23, 95% CI - 1.56 to - 0.91; I = 0% compared to the control group, indi- P = 0.00001, I = 19% meaning the patients cating that good glucose control was continu- experienced less time in the hypoglycemia ously maintained without requiring an excess of phase with this artiﬁcial pancreas as compared insulin (Fig. 2). to the control group (Fig. 2). Also, the numbers of hypoglycemic events Daily insulin required (24-h basis) also sig- were not signiﬁcantly different with WMD niﬁcantly favored artiﬁcial pancreas with WMD - 0.83, 95% CI - 1.76 to 0.10; P = 0.08, I =0% - 3.43, 95% CI - 4.27 to - 2.59; P = 0.00001, (Fig. 2). Diabetes Ther (2018) 9:1269–1277 1273 Fig. 2 Comparing artiﬁcial pancreas with the control group (part 1) Time spent outside the euglycemic phase P = 0.00001 meaning that patients using artiﬁ- (24 h-basis) also signiﬁcantly favored artiﬁcial cial pancreas hardly suffered any hyperglycemic pancreas with WMD - 6.28, 95% CI - 10.67 to stage as shown in Fig. 3. However, this result - 1.88; P = 0.005. This meant that most of the was also moderately heterogeneous. time, patients using artiﬁcial pancreas were in the euglycemic phase (neither experiencing DISCUSSION hypoglycemia nor hyperglycemia) as shown in Fig. 3. However, the results were moderately The use of a fully integrated artiﬁcial pancreas heterogeneous. in patients with T1DM was previously demon- Moreover, the time spent in the hyper- strated . glycemia phase (glucose [ 10.0 mmol/L) also The current results showed artiﬁcial pancreas signiﬁcantly favored artiﬁcial pancreas with to be signiﬁcantly more effective compared to WMD - 13.20, 95% CI - 16.47 to - 9.94; its control group in terms of glucose 1274 Diabetes Ther (2018) 9:1269–1277 Fig. 3 Comparing artiﬁcial pancreas with the control group (part 2) concentration, time spent in the hypoglycemic studied twice, once using their personal open- phase, and insulin delivery during a 24-h per- loop technique, and then a second time using iod. Artiﬁcial pancreas was also safer to use the closed-loop (artiﬁcial) system . Other owing to its association with a signiﬁcantly research further complemented the closed-loop insulin delivery technique [23, 24]. lower time period in the hyperglycemia phase, its signiﬁcant maintenance of a longer eug- Further improvement is being considered in lycemic period, and its lack of association with relation to this artiﬁcial pancreas . Also, any signiﬁcantly higher episode of hypo- useful tools have already been devised to glycemic event compared to its control. improve the assessment of glycemic variability Similarly, Hovorka et al. showed that artiﬁ- in patients with artiﬁcial pancreas . cial pancreas improved overnight control of glucose level and decreased the rate of noctur- LIMITATIONS nal hypoglycemia in patients with T1DM within a study time period of 3 months . This analysis also has limitations: (a) The Another multicenter study showed this artiﬁcial number of participants was extremely limited; pancreas to be very effective and safe to use in however, when compared to other previously patients with T1DM . Similarly, through a published studies, this analysis included a large multicenter 6-month trial of 24/7 automated number of patients. (b) The different follow-up insulin delivery in 2014, Kovatchev et al. time periods could have had an impact on the recently showed closed-loop control technology results obtained. (c) The range of the eug- to have matured and to appear safe for long- lycemic phase was supposed to be a glucose term use in patients with T1DM . level ranging between 3.9 and 8.0 mmol/L; This new device was even considered effec- however, a few studies recorded a glucose level tive in pediatric participants. Weinzimer et al. varying between 4.4 and 8.0 mmol/L or 3.9 to recently demonstrated fully automated closed- 10.0 mmol/L which might have contributed to loop insulin delivery versus semi-automated the moderate level of heterogeneity in this hybrid control in pediatric candidates with particular subgroup. (d) The inclusion of one T1DM . Insulin delivery using artiﬁcial non-randomized study might have introduced pancreas was further illustrated in the Virginia bias, contributing to the limitations in this experience, wherein the participants were analysis. (e) The control groups were not similar Diabetes Ther (2018) 9:1269–1277 1275 in all the studies, which might be another lim- Authorship Contributions. Xia Dai, Zu- itation of this analysis. (f) Utilizing sensor aug- chun Luo, Lu Zhai, Wen-piao Zhao, and Feng mented pump as the control group is the Huang were responsible for the conception and current clinical golden standard which artiﬁcial design, acquisition of data, analysis and inter- pancreas needs to be able to outperform if pretation of data, drafting the initial manu- clinically relevant. So, another limitation of this script and revising it critically for important study might be the lack of an analysis strictly intellectual content. Xia Dai wrote the ﬁnal dealing with studies comparing artiﬁcial pan- manuscript. All the authors approved the creas and sensor-augmented pump. However, manuscript as it is. the number of studies reporting this control was Disclosures. Xia Dai, Zu-chun Luo, Lu Zhai, too small. Wen-piao Zhao, and Feng Huang declare that they have nothing to disclose. They do not have CONCLUSION any personal, ﬁnancial, commercial, or aca- demic conﬂicts of interest. According to the results of this analysis, artiﬁ- cial pancreas might be considered an effective Compliance with Ethics Guidelines. This and safe alternative to be used during a 24-h meta-analysis is based on previously conducted basis in patients with T1DM. Several beneﬁts of studies and does not contain any studies with the artiﬁcial pancreas in maintaining and human participants or animals performed by improving glucose levels were observed in any of the authors. comparison to its control. Nevertheless, a major Data Availability. All data generated or shortcoming of this analysis is the extremely analyzed during this study are included in this limited number of patients analyzed. published article. Open Access. This article is distributed ACKNOWLEDGEMENTS under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/ Funding. This research was supported by by-nc/4.0/), which permits any noncommer- National Natural Science Foundation of China cial use, distribution, and reproduction in any (No. 81560046), Guangxi Natural Science medium, provided you give appropriate credit Foundation (No. 2016GXNSFAA380002), Sci- to the original author(s) and the source, provide entiﬁc Project of Guangxi Higher Education a link to the Creative Commons license, and (No. KY2015ZD028), Science Research and indicate if changes were made. Technology Development Project of Qingxiu District of Nanning (No. 2016058), and Lisheng Health Foundation pilotage fund of Peking (No. LHJJ20158126). No funding or sponsorship was REFERENCES received for the publication of this article. The article processing charges were funded by the 1. Zayed H. Genetic epidemiology of type 1 diabetes authors. in the 22 Arab countries. Curr Diab Rep. 2016;16(5):37. Authorship. All named authors meet the 2. Hurren KM, O’Neill JL. Pharmacodynamic and International Committee of Medical Journal pharmacokinetic evaluation of insulin glargine Editors (ICMJE) criteria for authorship for this U300 for the treatment of type 1 diabetes. Expert article, take responsibility for the integrity of Opin Drug Metab Toxicol. 2016;12(12):1521–6. the work as a whole, and have given their 3. McEwan P, Bennett H, Fellows J, Priaulx J, Bergen- approval for this version to be published. heim K. The health economic value of changes in 1276 Diabetes Ther (2018) 9:1269–1277 glycaemic control, weight and rates of hypogly- pump therapy in adults with type 1 diabetes: a caemia in type 1 diabetes mellitus. PLoS One. multicentre randomised crossover trial. Lancet. 2016;11(9):e0162441. 2016. 4. Ramirez-Rincon A, Hincapie-Garcı´a J, Arango CM, 15. Russell SJ, El-Khatib FH, Sinha M, et al. Outpatient et al. Clinical outcomes after 1 year of augmented glycemic control with a bionic pancreas in type 1 insulin pump therapy in patients with diabetes in a diabetes. N Engl J Med. 2014;371(4):313–25. specialized diabetes center in Medellı´n, Colombia. Diabetes Technol Ther. 2016;18(11):713–8. 16. Russell SJ, Hillard MA, Balliro C, Magyar KL, et al. Day and night glycaemic control with a bionic 5. Voelker R. ‘‘Artiﬁcial pancreas’’ is approved. JAMA. pancreas versus conventional insulin pump therapy 2016;316(19):1957. in preadolescent children with type 1 diabetes: a randomised crossover trial. Lancet Diabetes Endo- 6. Higgins JPT, Altman DG. Chapter 8: Assessing risk crinol. 2016;4(3):233–43. of bias in included studies. In: Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews 17. Breton M, Farret A, Bruttomesso D, et al. Fully of interventions. Chichester: Wiley; 2008. integrated artiﬁcial pancreas in type 1 diabetes: p. 187–241. modular closed-loop glucose control maintains near normoglycemia. Diabetes. 2012;61(9):2230–7. 7. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and 18. Hovorka R, Kumareswaran K, Harris J, et al. Over- meta-analyses of studies that evaluate healthcare night closed loop insulin delivery (artiﬁcial pan- interventions: explanation and elaboration. BMJ. creas) in adults with type 1 diabetes: crossover 2009;339:b2700. randomised controlled studies. BMJ. 2011;13(342):d1855. 8. Sterne JA, Egger M, Smith GD. Systematic reviews in health care: investigating and dealing with publi- 19. Del Favero S, Place J, Kropff J, et al. Multicenter cation and other biases in meta-analysis. BMJ. outpatient dinner/overnight reduction of hypo- 2001;323(7304):101–5. glycemia and increased time of glucose in target with a wearable artiﬁcial pancreas using modular 9. Blauw H, van Bon AC, Koops R, DeVries JH. Per- model predictive control in adults with type 1 dia- formance and safety of an integrated bihormonal betes. Diabetes Obes Metab. 2015;17(5):468–76. artiﬁcial pancreas for fully automated glucose con- trol at home. Diabetes Obes Metab. 20. Kovatchev B, Cheng P, Anderson SM, et al. Feasi- 2016;18(7):671–7. bility of long-term closed-loop control: a multi- center 6-month trial of 24/7 automated insulin 10. Kropff J, Del Favero S, Place J, et al. 2 month eve- delivery. Diabetes Technol Ther. 2017;19(1):18–24. ning and night closed-loop glucose control in patients with type 1 diabetes under free-living 21. Weinzimer SA, Steil GM, Swan KL, et al. Fully conditions: a randomised crossover trial. Lancet automated closed-loopinsulin delivery versus Diabetes Endocrinol. 2015;3(12):939–47. semiautomated hybrid control in pediatric patients with type1 diabetes using an artiﬁcial pancreas. 11. Renard E, Farret A, Kropff J, et al. Day-and-night Diabetes Care. 2008;31(5):934–9 closed-loop glucose control in patients with type 1 diabetes under free-living conditions: results of a 22. Clarke WL, Anderson S, Breton M, et al. Closed- single-arm 1-month experience compared with a loopartiﬁcial pancreas using subcutaneous glucose previously reported feasibility study of evening and sensing and insulin delivery anda model predictive night at home. Diabetes Care. 2016;39(7):1151–60. control algorithm: the Virginia experience. J Dia- betes SciTechnol. 20091;3(5):1031–8. 12. Thabit H, Lubina-Solomon A, Stadler M, et al. Home use of closed-loop insulin delivery for over- 23. Shariﬁ A, De Bock MI, Jayawardene D, night glucose control in adults with type 1 diabetes: et al.Glycemia, treatment satisfaction, cogni- a 4-week, multicentre, randomised crossover study. tion,and sleep quality in adults and adolescents Lancet Diabetes Endocrinol. 2014;2(9):701–9. with type 1 diabetes when using a closed-loop sys- tem overnight versus sensor-augmented pump with 13. Kovatchev BP, Renard E, Cobelli C, et al. Safety of low-glucosesuspend function: a randomized cross- outpatient closed-loop control: ﬁrst randomized over study. Diabetes Technol Ther. crossover trials of a wearable artiﬁcial pancreas. 2016;18(12):772–83. Diabetes Care. 2014;37(7):1789–96. 24. Weisman A, Bai JW, Cardinez M, Kramer CK, Per- 14. El-Khatib FH, Balliro C, Hillard MA, et al. Home use kins BA. Effect of artiﬁcialpancreas systems on gly- of a bihormonal bionic pancreas versus insulin caemic control in patients with type 1 diabetes: a Diabetes Ther (2018) 9:1269–1277 1277 systematicreview and meta-analysis of outpatient 26. Garcia A, Balo AK, Buckingham BA, Hirsch IB, randomised controlled trials. Lancet DiabetesEn- Peyser TA. Application ofglycemic variability per- docrinol. 2017;5(7):501–12 centage: implications for continuous glucose mon- itorutilization and analysis of artiﬁcial pancreas 25. Gildersleeve R, Riggs SL, Chern ˜avvsky DR, Breton data. Diabetes Technol Ther. 2017;19(12):699–706. MD, DeBoer MD. Improvingthe safety and func- tionality of an artiﬁcial pancreas system for use in youngerchildren: input from parents and physi- cians. Diabetes Technol Ther. 2017;19(11):660–74
Diabetes Therapy – Springer Journals
Published: May 9, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera