Insulin and oral agents for managing cystic fibrosis-related diabetes

  • Conclusions changed
  • Review
  • Intervention

Authors


Abstract

Background

The Cystic Fibrosis Foundation recommends both short-term and long-acting insulin therapy when cystic fibrosis-related diabetes has been diagnosed. Diagnosis is based on: an elevated fasting blood glucose level greater than 6.94 mmol/liter (125 mg/deciliter); or symptomatic diabetes for random glucose levels greater than 11.11 mmol/liter (200 mg/deciliter); or glycated hemoglobin levels of at least 6.5%.

Objectives

To establish the effectiveness of agents for managing diabetes in people with cystic fibrosis in relation to blood sugar levels, lung function and weight management.

Search methods

We searched the Cochrane Cystic Fibrosis and Genetic Disorders Group Trials Register comprising references identified from comprehensive electronic database searches and handsearches of relevant journals and abstract books of conference proceedings. We also handsearched abstracts from pulmonary symposia and the North American Cystic Fibrosis Conferences.

Date of the most recent search of the Group's Cystic Fibrosis Trials Register: 22 July 2013.

Selection criteria

Randomized controlled trials comparing all methods of diabetes therapy in people with diagnosed cystic fibrosis-related diabetes.

Data collection and analysis

Two authors independently extracted data and assessed the risk of bias in the included studies.

Main results

The searches identified 19 studies (28 references). Three studies (107 participants) are included: one comparing insulin with oral repaglinide and no medication (short-term single-center study of seven patients with cystic fibrosis-related diabetes and normal fasting glucose); one comparing insulin with oral repaglinide and placebo (long-term multi-center study with 81 patients, 61 of whom had cystic fibrosis-related diabetes); and one 12-week single-center study comparing the long-acting insulin, glargine to short-term neutral protamine Hagedorn insulin. The long-term trial of insulin and repaglinide demonstrated no significant difference between treatments. In the smaller study comparing insulin and oral repaglinide, there were two incidents of significant hypoglycemia in the insulin group compared to one in the repaglinide group; in the larger study there were five incidents of significant hypoglycemia in the insulin group and six in the repaglinide group. The study comparing glargine to neutral protamine Hagedorn insulin demonstrated a statistically non-significant weight increase in with longer-acting insulin given at bedtime and reported a mean of six hypoglycemia events in the glargine group compared to five events in the neutral protamine Hagedorn insulin group. None of the three included studies were powered to show a significant improvement in lung function.

Authors' conclusions

This review has not found any significant conclusive evidence that long-acting insulins, short-acting insulins or oral hypoglycemic agents have a distinct advantage over one another in controlling hyperglycemia or clinical outcomes associated with cystic fibrosis-related diabetes. While some cystic fibrosis centers use oral medications to help control diabetes, the Cystic Fibrosis Foundation (USA) clinical practice guidelines support the use of insulin therapy and this remains the most widely-used treatment method. Randomized controlled trials specifically related to controlling diabetes with this impact on the course of pulmonary disease process in cystic fibrosis continue to be a high priority.

There is no demonstrated advantage yet established for using oral hypoglycemic agents over insulin, and further studies need to be evaluated to establish whether there is clear benefit for using hypoglycemic agents. Agents that potentiate insulin action, especially agents with additional anti-inflammatory potential should be further investigated to see if there may be a clinical advantage to adding these medications to insulin as adjuvant therapy.

Résumé scientifique

Insuline et agents oraux pour la prise en charge du diabète lié à la mucoviscidose

Contexte

La Cystic Fibrosis Foundation recommande un traitement à la fois à l'insuline à action rapide et à action prolongée lorsqu'un diabète lié à la mucoviscidose a été diagnostiqué. Le diagnostic se fonde sur : un taux de glucose dans le sang à jeun élevé, supérieur à 6,94 mmol/litre (125 mg/décilitre) ; ou un diabète symptomatique pour des taux de glucose aléatoires supérieurs à 11,11 mmol/litre (200 mg/décilitre) ou des taux d'hémoglobine glyquée d'au moins 6,5 %.

Objectifs

Établir l'efficacité d'agents pour la prise en charge du diabète chez les personnes atteintes de mucoviscidose en termes de taux de sucre dans le sang, de fonction pulmonaire et de gestion du poids.

Stratégie de recherche documentaire

Nous avons effectué des recherches dans le registre d'essais cliniques du groupe Cochrane sur la mucoviscidose et autres maladies génétiques constitué de références identifiées lors de recherches exhaustives dans les bases de données électroniques et de recherches manuelles de revues pertinentes et de résumés d'actes de conférence. Nous avons également effectué des recherches manuelles dans les actes de symposiums sur les maladies pulmonaires et les actes des North American Cystic Fibrosis Conferences.

Date de la recherche la plus récente effectuée dans le registre des essais cliniques du groupe sur la mucoviscidose : le 22 juillet 2013.

Critères de sélection

Les essais contrôlés randomisés comparant toutes les méthodes de traitement du diabète chez les personnes ayant reçu un diagnostic de diabète lié à la mucoviscidose.

Recueil et analyse des données

Deux auteurs ont indépendamment extrait des données et évalué les risques de biais dans les études incluses.

Résultats principaux

Les recherches ont produit 19 études (28 références). Trois études (soit 107 participants) sont incluses : une comparant l'insuline à la répaglinide orale et à l'absence de médicament (étude monocentrique à court terme de sept patients atteints d'un diabète lié à la mucoviscidose et ayant un taux de glucose à jeun normal) ; une comparant l'insuline à la répaglinide orale et à un placebo (étude multicentrique à long terme sur 81 patients, dont 61 atteints de diabète lié à la mucoviscidose) et une étude monocentrique de 12 semaines comparant l'insuline à action prolongée, la glargine, à l'insuline neutre protamine Hagedorn à action rapide. L'essai à long terme de l'insuline et de la répaglinide n'a démontré aucune différence significative entre les traitements. Dans l'étude plus petite comparant l'insuline et la répaglinide orale, on a observé deux incidents d'hypoglycémie significative dans le groupe avec insuline comparé à un dans le groupe avec la répaglinide ; dans l'étude à plus grande échelle, on a observé cinq incidents d'hypoglycémie significative dans le groupe avec insuline et six dans le groupe avec répaglinide. L'étude comparant la glargine à l'insuline neutre protamine Hagedorn a démontré une augmentation de poids non-significative statistiquement avec l'insuline à action prolongée administrée au moment du coucher et a rapporté une moyenne de six événements d'hypoglycémie dans le groupe avec glargine comparé à cinq événements dans le groupe avec insuline neutre protamine Hagedorn. Aucune des trois études incluses ne disposait de la puissance nécessaire pour montrer une amélioration significative de la fonction pulmonaire.

Conclusions des auteurs

Cette revue n'a trouvé aucune preuve concluante significative indiquant que les insulines à action prolongée, les insulines à action rapide ou les agents hypoglycémiques oraux présentaient un avantage distinct les uns par rapport aux autres en termes de régulation de l'hyperglycémie ou de résultats cliniques associés au diabète lié à la mucoviscidose. Si certains centres de lutte contre la mucoviscidose utilisent des médicaments oraux pour aider à contrôler le diabète, les directives de pratique clinique de la Cystic Fibrosis Foundation (Etats-Unis) soutiennent l'utilisation de l'insulinothérapie et cela reste la méthode de traitement la plus couramment utilisée. Les essais contrôlés randomisés concernant spécifiquement le contrôle du diabète avec cet impact sur l'évolution du processus de la maladie pulmonaire dans la mucoviscidose demeurent une priorité majeure.

Aucun avantage démontré n'a encore été établi concernant l'utilisation d'agents hypoglycémiques oraux par rapport à l'insuline et des études supplémentaires doivent être évaluées pour déterminer si l'utilisation d'agents hypoglycémiques apporte un bénéfice évident. Les agents qui potentialisent l'action de l'insuline, en particulier les agents ayant un potentiel anti-inflammatoire supplémentaire, doivent être étudiés davantage pour déterminer s'il peut y avoir un avantage clinique à ajouter ces médicaments à l'insuline au titre de traitement d'appoint.

Plain language summary

Insulin and oral agents for managing cystic-fibrosis related diabetes

Cystic fibrosis is a serious genetic disorder which damages the lungs and pancreas. It is the most common life-limiting inherited disease in white populations. The pancreas produces insulin which is a hormone needed by many cells in the body (specifically in the liver, muscle and fat) to take sugar into the cell and convert it into energy. People with cystic fibrosis need high-calorie diets to maintain enough muscles to compensate for breathing difficulties resulting from lung damage. It is therefore important for people with diabetes as an additional complication in cystic fibrosis to turn sugar into energy in the most efficient way, so they can manage breathing difficulties and maintain an ideal body weight. The inflammatory processes in cystic fibrosis can impair insulin production and diminish its effect by causing insulin resistance. In these situations, either artificial sources of insulin (such as long-acting glargine or short-acting protamine insulin) or other medications to enhance the release or action of the patient’s own insulin are needed to prevent a further decline in health.

We have included three randomized studies in this review. Two studies compared insulin with oral repaglinide and no medication or placebo and one study compared the long-acting insulin, glargine, to short-term neutral protamine Hagedorn insulin. We were not able to show that any of the treatments were better than the others. However, the amounts of insulin and repaglinide received were not comparable and this may lead to bias in the results. Only a few cases of hypoglycemia (low blood sugar) were reported in all of these studies, but these events resolved without further treatment. Longer-term studies are still needed which look at the effect on lung function of controlling cystic fibrosis-related diabetes, or treating pre-diabetes. There also needs to be research into the use of agents to be used together with insulin to enhance its action, especially those agents with additional anti-inflammatory potential.

Résumé simplifié

Insuline et agents oraux pour la prise en charge du diabète lié à la mucoviscidose

La mucoviscidose est un trouble génétique grave qui endommage les poumons et le pancréas. Il s'agit de la maladie héréditaire limitant l'espérance de vie la plus courante dans les populations blanches. Le pancréas produit de l'insuline qui est une hormone dont de nombreuses cellules du corps ont besoin (particulièrement dans le foie, les muscles et la graisse) pour absorber le sucre et le convertir en énergie. Les personnes atteintes de mucoviscidose ont besoin de régimes alimentaires riches en calories pour conserver suffisamment de muscles pour compenser les difficultés respiratoires dues aux lésions pulmonaires. Il est donc important pour les personnes souffrant d'un diabète représentant une complication supplémentaire de la mucoviscidose de convertir le sucre en énergie de la manière la plus efficace, de sorte qu'elles puissent faire face aux difficultés respiratoires et maintenir un poids idéal. Les processus inflammatoires dans la mucoviscidose peuvent diminuer la production d'insuline et réduire son effet en provoquant une résistance à l'insuline. Dans ce cas, des sources d'insuline artificielles (telles que l'insuline glargine à action prolongée ou l'insuline protamine à action rapide) ou d'autres médicaments visant à améliorer la libération ou l'action de l'insuline du patient sont nécessaires pour empêcher un déclin supplémentaire de la santé.

Nous avons inclus trois études randomisées dans cette revue. Deux études comparaient l'insuline à la répaglinide orale et à l'absence de médicament ou à un placebo et une étude comparait l'insuline à action prolongée, la glargine, à l'insuline neutre protamine Hagedorn à action rapide. Nous n'avons pas été en mesure de montrer que l'un des traitements était plus efficace que les autres. Cependant, les quantités d'insuline et de répaglinide reçues n'étaient pas comparables et cela peut conduire à un biais dans les résultats. Seuls quelques cas d'hypoglycémie (faible taux de sucre dans le sang) ont été rapportés dans l'ensemble des études, mais ces événements se sont résolus sans traitement supplémentaire. Il faut encore réaliser des études à plus long terme examinant l'effet sur la fonction pulmonaire du contrôle du diabète lié à la mucoviscidose ou du traitement du prédiabète. Des recherches sont également nécessaires concernant le recours à des agents à utiliser avec l'insuline pour améliorer son action, en particulier ceux ayant un potentiel anti-inflammatoire supplémentaire.

Notes de traduction

Traduit par: French Cochrane Centre 4th September, 2013
Traduction financée par: Pour la France : Minist�re de la Sant�. Pour le Canada : Instituts de recherche en sant� du Canada, minist�re de la Sant� du Qu�bec, Fonds de recherche de Qu�bec-Sant� et Institut national d'excellence en sant� et en services sociaux.

Background

Description of the condition

Survival of people with cystic fibrosis (CF) now extends well into adulthood, and as a consequence secondary disease processes are increasingly being recognized. Cystic fibrosis-related diabetes (CFRD) represents the highest secondary prevalence in CF and is seen in nearly 20% of adolescents and 40% to 50% of adults with CF (Moran 2009). The diagnosis of CFRD is currently made during a period of stable baseline health according to standard American Diabetic Association (ADA) criteria. This diagnosis is therefore made if the two-hour oral glucose tolerance plasma glucose is elevated (greater than or equal to 200 mg/dl (11.1 mmol/l)); there is a fasting plasma glucose greater than or equal to 126; A1C is greater than or equal to 6.5%; or if polyuria or polydipsia are in the presence of a casual glucose level of greater than or equal to 200 mg/dl. The diagnosis of CFRD can additionally be considered in CF patients with acute illness when they maintain fasting plasma glucose elevations, or two-hour post-prandial plasma glucose levels greater than or equal to 200 mg/dl for more than 48 hours (Moran 2010). There is a 36% occurrence of microvascular complications that includes kidney disease and retinopathy in CFRD, compared to 18% in the non-CF diabetic population (Landers 1997; Schwarzenberg 2007). Microvascular complications result from prolonged periods of hyperglycemia which inflict pathological changes within small blood vessels. These factors may impact on a reported 60% survival to age 30 years for individuals with CF but without diabetes, compared to 25% for individuals with CF and diabetes (Finkelstein 1988). The negative impact of diabetes has been more specifically linked to rapid decline in pulmonary function and body mass index (BMI) in people with CF (Lanng 1994), and most recently, a link to the quality of diabetes management has been reported (Moran 2010). These findings have important implications, as end-stage lung disease is the ultimate event leading to mortality in nearly all individuals with CF, and the added complication of diabetes may hasten this process, as well as the additional impacts on secondary disease processes as a consequence of improved survival of CF patients.

Description of the intervention

The Cystic Fibrosis Foundation guidelines recommend insulin as the treatment of choice, based on evidence supported by observation of increasing trends toward insulin deficiency over time (Moran 1991; Moran 2010); however, insulin deficiency is seldom absolute, as ketoacidosis is a very uncommon complication in CF. Increasing insulin resistance has further been correlated to the progressive development of impaired glucose tolerance seen in CF (Hardin 1997). The implication of both a relative decrease in pancreatic insulin release, and an increase in insulin resistance to glucose metabolism by the body suggests potential roles for both oral insulin-releasing hypoglycemic and oral insulin-sensitizing medications in the management of CFRD insulin therapy.

How the intervention might work

Therapeutic reports for CFRD which have discussed the impact on pulmonary function and body weight (as a proportion to the square of the height known as the BMI) have been limited to individuals managed by insulin therapy. The institution of insulin therapy aimed toward optimizing glycemic control to the range recommended by the ADA (ADA 2004) has been recently demonstrated to have a positive impact toward improving glycemic control based on HbA1c measurements and to minimize long-term micro-vascular disease based on correlations from a recent prognostic cohort study (Schwarzenberg 2007).The effect of maintaining HbA1c at or below 7% in CFRD is now significantly associated with minimizing the complications of diabetes on microvascular disease at a prevalence even more reduced than is seen for type 1 and type 2 diabetes.

Why it is important to do this review

Reports of the use of oral hypoglycemic agents for controlling HbA1c are limited. There is only one case-based published study reporting the effects of insulin-sensitizing agents (Onady 2006a). Recommendations for glycemic control by the ADA emphasize the importance of optimizing the secondary prevention of the complication of microvascular disease in type 1 and type 2 diabetes, but there is still uncertainty as to how this control impacts on microvascular disease in CFRD, pulmonary function or optimal weight through improved carbohydrate metabolism (ADA 2004).

Objectives

To establish the effectiveness of agents for managing diabetes in people with cystic fibrosis in relation to blood sugar levels, lung function and weight management.

Methods

Criteria for considering studies for this review

Types of studies

Randomized controlled trials.

Types of participants

Participants of all ages must be established as having CF, as determined by a positive diagnostic test for CF that may include sweat, nasal epithelial and genotype testing associated with pulmonary or gastrointestinal disease or both. Diabetes from this cohort will be further established by oral glucose tolerance testing, A1C data, two fasting or random blood sugars as defined by ADA standards (ADA 2004). The CF Foundation Guidelines (USA) related to a diagnosis during pulmonary exacerbations and in those with CF requiring feeding tubes may also be additionally used in establishing diagnostic criteria for CFRD.

Types of interventions

We aimed to compare different insulin regimens and regimens of oral diabetic medications to each other. We originally did not plan to include studies that compared an active treatment-only regimen with a placebo control arm, as we thought therapy for diabetes was unlikely to be withheld on ethical grounds. However, we did identify such studies and included them and note this post hoc change was made to the review. We planned to include studies where insulin regimens plus either an oral medication or placebo are compared. In this analysis we aimed to consider studies comparing different classes of these therapeutic agents, as well as assessing strategies to reduce side effects and harm in using these agents.

Insulin preparations to be considered consist of:

  1. short-acting insulin, with duration of action typically lasting two to four hours;

  2. intermediate-acting insulin, with duration of action typically to 12 hours;

  3. long-acting insulin, with duration of actions approaching 24 hours.

Post hoc change

Oral anti-diabetic agents to be considered are:

  1. sulphonyureas and related agents;

  2. biguanides and related agents;

  3. glitazones and related agents;

  4. other agents that specifically manage hyperglycaemia.

Types of outcome measures

Primary outcomes
  1. Biochemical measures of glycaemic control, ie HbA1c, fasting and two-hour post-meal serum blood sugar values

  2. Pulmonary function (eg forced expiratory volume (FEV1) and forced vital capacity (FVC))

  3. Assessment of nutritional status (eg body mass index (BMI))

Secondary outcomes
  1. Prevalence of microvascular and macrovascular disease

    1. retinopathy

    2. neuropathy

    3. nephropathy

  2. Rate of pulmonary exacerbations [post hoc change]

  3. Complications of therapeutic management

    1. hypoglycemia (specifically related to oral hypoglycemic and insulin agents)

    2. liver toxicity (specifically related to the thiazolidinedione class)

    3. metabolic effects on acid-base status (specifically related to biguanides (metformin))

  4. Clinical status (post hoc change)

    1. six-minute walk test

    2. health-related quality of life (HRQoL) instrument (eg the Cystic Fibrosis Questionnaire-Revised (CFQ-R) (Quittner 2009))

    3. mortality

Search methods for identification of studies

Electronic searches

Relevant trials were sought from the Group's Cystic Fibrosis Trials Register using the terms: insulin OR diabetes.

The Cystic Fibrosis Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of The Cochrane Library), quarterly searches of MEDLINE, a search of EMBASE to 1995 and the prospective handsearching of two journals - Pediatric Pulmonology and the Journal of Cystic Fibrosis. Unpublished work is identified by searching the abstract books of three major cystic fibrosis conferences: the International Cystic Fibrosis Conference; the European Cystic Fibrosis Conference and the North American Cystic Fibrosis Conference. For full details of all searching activities for the register, please see the relevant sections of the Cystic Fibrosis and Genetic Disorders Group Module.

Date of the most recent search of the Group's CF Trials Register: 22 July 2013.

In December 2010 we searched PubMed for articles published in 2009 and 2010 using the terms cystic fibrosis AND diabetes AND insulin with Limits.

Searching other resources

We also handsearched abstracts from pulmonary and North American Cystic Fibrosis Conference symposia up to 2006 for the original version of the review.

Data collection and analysis

Selection of studies

Two authors independently selected trials to be included in the review. No disagreement occurred between the authors, but if in future these do occur the authors will resolve these by discussion.

Data extraction and management

Each author independently extracted data using standard data acquisition forms; however, the authors judged the data from one of the included cross-over studies to have been more appropriately analysed in the original paper and since they were not able to obtain first-arm data from the study investigators, they have reported results narratively directly from the publication (Moran 2001). If it is necessary for future updates of the review, the review authors will request individual or additional summary data from the trial authors.

We planned to measure outcome data at 1, 3, 6, 12 months and annually thereafter. We were able to report data for glucose levels, weight, fat mass and lean mass and hypoglycemic events at three months and for BMI, FEV1 and FVC at 12 months. If for future updates, if trial investigators report any other time points relevant and applied toward the primary and secondary outcomes, then we will give consideration to these too.

For any data presented as means and standard errors in the cross-over trial by Grover and the parallel trial by Moran (Grover 2008; Moran 2009), we converted the standard errors (SEs) to standard deviations (SDs) to allow data to be entered in the analysis. We did not combine data from the cross-over study with any data from the parallel study.

Assessment of risk of bias in included studies

In order to establish a risk of bias for the included trial, for the original review each author assessed the methodological quality of the trial. In future, authors will also monitor the consistency of inclusion and exclusion criteria, the appropriateness of comparisons and the rate of attrition.

For the updates from 2013 onwards, the authors assessed the risk of bias of the included trials according to the methods recommended by the Cochrane Collaboration (Higgins 2011). These methods examine internal validity by addressing selection, performance and attrition bias. Authors also assessed concealment of treatment allocation using this methodology. Authors prioritized external validity to assure the availability of adequate information regarding the characteristics of participants and their results in the included trials. Authors also compared the outcomes listed in the methods section of the study publication to the results reported to assess for selective outcome reporting. For each included trial, the authors gave a judgement of low, high or unclear risk of bias for each of the criteria examined.

Measures of treatment effect

In future updates of this review, for binary outcomes the goal will be to calculate a pooled estimate of treatment effect for each outcome across studies using the risk ratios (RRs) and their 95% confidence intervals (CIs).

For continuous outcomes, the authors measured the mean change from baseline for each group or mean post-treatment or post-intervention values and SD for each group. Where it was necessary, we calculated the SD from the SE which was presented in the trial publication. The authors used a pooled estimate of treatment effect by calculating the mean difference (MD) with 95% CIs.

Unit of analysis issues

Cross-over trials may be relevant when both insulin and oral agents are involved and indeed two of the included trials were of cross-over design (Grover 2008; Moran 2001). The authors treated the data from the Grover trial as if it had been a parallel trial since the first-arm study data were not available as confirmed by the study authors (Grover 2008). No first-arm data were available for analysis from the Moran study despite the review authors contacting the study authors for more information, and on previous advice from a statistician the results from this study, which were analysed more appropriately in the original study publication, are reported in the review narratively (Moran 2001). The authors did not combine data from the cross-over study with those from the parallel study. If, in the future, the authors include further cross-over trials, they will decide whether to report these results narratively also, or they may decide to analyse the cross-over trials as if they had been parallel trials (assuming a correlation of zero as the most conservative estimate). Elbourne says that this approach produces conservative results as it does not take into account within-patient correlation. Also each participant appears in both the treatment and control group, so the two groups are not independent (Elbourne 2002).

Dealing with missing data

For this version of the review, the authors have presented the data which are reported in the study publications. They tried to obtain first-arm data from the authors of the 2001 Moran study and from the Grover study, but the investigators informed the authors that it was not possible to provide these data as they were no longer available (Grover 2008; Moran 2001). In future updates of this review, the authors will seek data on the number of participants with each outcome measure by allocated treated group, irrespective of compliance or otherwise excluded from treatment at follow up. If these data are not available in publications, the authors will make every attempt to contact the trialists for further information.

Assessment of heterogeneity

When the authors are able to include a sufficient number of trials in the review, they will test for heterogeneity between trial results using a standard chi-squared test and the I2 statistic (Higgins 2003). This measure describes the percentage of total variation across studies that are due to heterogeneity rather than by chance. The values of I2 lie between 0% and 100%, and a simplified categorization of heterogeneity that we plan to use is of low (I2 value of 25%), moderate (I2 value of 50%), and high (I2 value of 75%) (Higgins 2003).

Data synthesis

The authors have analysed data using the fixed-effect model and will continue to do so for future updates, unless additional data result in statistical heterogeneity (I2 value of over 50%) in which case they will use a random-effects model when combining data from trials.

Subgroup analysis and investigation of heterogeneity

In future, if sufficient studies (n = 10) are included and there is significant heterogeneity between the studies, the authors plan to perform subgroup analyses investigating:

  • age (pediatric versus adult patients (0 to 18 years versus over 18 years of age);

  • severity of baseline pulmonary function with mild disease (down to an FEV1 of 80% predicted), moderate disease (between 80% and 40% predicted) and severe disease (less than 40% predicted);

  • stratification of various levels glycemic control as identified through HbA1c with adequate control at less than 7%, intermediate control between 7% to 8% and in poor control greater than 8%.

'End-of-life' onset of diabetes may imply a different quality of diabetic control than for the younger diabetic identified earlier and with a milder baseline chronic pulmonary disease setting.

Sensitivity analysis

When we are able to include sufficient trials we plan to perform a sensitivity analysis based on the potential diversity in study parameters. As an example, many studies from North America omit correlations of diabetic control to HbA1c; whereas European studies tend to follow this closely. Plotting effect estimates against risk of bias measures or performing cumulative meta-analysis based on quality order may also be effective at addressing problems surrounding composite scales.

Results

Description of studies

Results of the search

The searches identified 19 studies (28 references).

Included studies

Three trials were included in this review (Grover 2008; Moran 2001; Moran 2009).

Trial Design

Two trials were of cross-over design (Grover 2008; Moran 2001) and one was of parallel design (Moran 2009). Both cross-over trials included a washout period. One of these studies utilized three separate single doses over a one-month to two-month period (Moran 2001). The remaining study used a one month washout period (Grover 2008). Duration of the trials ranged from a single-dose study (Moran 2001) to 12 months (Moran 2009). Only one included study was multicenter (14 centres in the USA, Canada and UK) (Moran 2009).

Participants

The number of participants enrolled in the trials ranged from 7 (Moran 2001) to 74 (Moran 2009). However, the proportion of withdrawals ranged from 0% (Moran 2001) to 27% (Moran 2009) meaning the number of participants for whom data are available ranges from 7 (Moran 2001) to 61 (Moran 2009). In all three included studies there were roughly equal numbers of male and female participants enrolled overall. Mean (SD) age ranged from 24 (5) years (Moran 2001) to 34 (8) years (Grover 2008). All studies stated that participants were clinically well and two studies required that participants had not experienced acute illness in the previous two (Moran 2009) or three months (Moran 2001).

Interventions

The three included studies employed different interventions.

Two compared insulin with oral repaglinide (Moran 2001; Moran 2009). One of these studies had a third 'no medication' arm (Moran 2001), while the second study by Moran included a third placebo arm (Moran 2009). The earlier study considered single-dose treatments given in random order of no medication versus insulin (0.1 unit/kg) versus repaglinide (1 mg); both insulin and repaglinide were administered 10 minutes prior to a meal (Moran 2001). This study was mostly used to develop more a clinically relevant study (Moran 2009), and to date the only evidence-based study comparing insulin to an oral alternative. The second study compared insulin (0.5 unit insulin aspart/15 g dietary carbohydrate) to repaglinide (2 mg) three times daily before meals (time frame not stated) (Moran 2009). In this second study by Moran, 17% of participants had doses of repaglinide reduced due to hypoglycemia (Moran 2009).

The third study compared neutral protamine Hagedorn (NPH) insulin to glargine; both treatments were administered once daily at bedtime (Grover 2008). The average total daily insulin dose observed was 0.7 units/kg/day for each insulin, with mean (SD) insulin to carbohydrate ratios of 1.5 (0.2) NPH and 1.3 (0.1) glargine aspart/15 g carbohydrate, P = 0.05. The mean (SD) dose of glargine was 46 (4)% and of NPH 38 (3)% of the total daily insulin dose (Grover 2008).

Outcome measures

All three studies reported on blood glucose levels and adverse events (Grover 2008; Moran 2001; Moran 2009). The two longer studies also reported quality of life (QoL) along with measures of nutritional status including body mass index (BMI), weight, fat and lean mass (Grover 2008; Moran 2009). Only the second Moran trial reported on lung function and National Institute for Health (NIH) prognostic score (Moran 2009).

Excluded studies

Sixteen studies identified in the searches were excluded. Thirteen studies were not randomized controlled studies (Borowitz 2005; Chernoff 2002; Franzese 2005; Hardin 2009; König 2005; Mahroukh 2005; Marshall 2005; Milla 2005; Onady 2006; Peraldo 1998; Reali 2006; Sulli 2007; Ward 1999). One study was of type 1 diabetes not CFRD (Teeter 2004); one study included participants with glucose intolerance and not CFRD (Minicucci 2008); and one study was of interventions to promote treatment adherence in children with type 1 diabetes or CF and not a comparison of treatments for CFRD (Driscoll 2009).

Risk of bias in included studies

Allocation

Two studies were judged to have an unclear risk of bias for both sequence generation and allocation concealment (Grover 2008; Moran 2001). Grover did not provide details of either the randomization sequence or how this was concealed. A statement was made that groups were similar in demographic characteristics and gender proportion, but a table of baseline comparisons was not provided (Grover 2008). In the earlier study, Moran does state that the interventions were given in random order, but does not describe how this order was generated or whether it was concealed (Moran 2001).

In the later Moran study, the paper stated that block randomization using a pseudo-random number generator with stratification by center was employed; so we judged this study to have a low risk of bias for sequence generation. However, allocation concealment was not discussed in the paper, leading to an unclear risk of bias (Moran 2009).

Blinding

We estimate the risk of bias from blinding as high in two studies (Grover 2008; Moran 2001). The earlier Moran paper does not discuss blinding of participants, clinicians or outcome assessors, however, as the interventions were either no medication, insulin via a syringe or oral repaglinide it would have been impossible to blind the participants or the clinicians to the treatment group (Moran 2001). Similarly, the Grover article states that this was a cross-over study so that once randomized to start either therapy wing (bedtime NPH-Aspart OR bedtime glargine) the patient (and physician) would be aware of the treatment given by a factor of the frequency of dosing (Grover 2008). 

In the later Moran study, blinding was done for patients receiving either the oral agent or the oral placebo, which leads to a low risk of bias (Moran 2009). However, it should be noted that it was not possible to blind patients receiving insulin, so this arm of the trial is at risk of bias from blinding (Moran 2009).

Incomplete outcome data

We judged one study to be at high risk of bias from incomplete outcome data (Moran 2001). In this study, while there was no attrition over this brief five-hour study period, the paper stated outcomes were measured at 20-minute intervals, but only presented data for two and five-hour time-points.

We judged the Grover study to have an unclear risk of bias as it was reported that one male participant dropped out of the group who received NPH first, however, no reason was given for this and it was also not stated in which arm of the trial the drop out occurred (Grover 2008).

We judged the later Moran study to have a low risk of bias (Moran 2009). All individuals dropping out from the study were detailed and there were no significant differences in baseline characteristics for individuals dropping out, with the exception of the CFQOL measure for which it was noted that those stopping early had lower eating disturbance scores and social/marginalization scales (P <0.05) (Moran 2009).

Selective reporting

We judged all three included studies to have a high risk of bias due to selective reporting (Grover 2008; Moran 2001; Moran 2009). Although all three mention the all outcomes stated in the 'Methods' section in the 'Results' sections, data are not always provided, often there are just statements that there was no difference between treatment groups. We contacted the authors for the data to enter in our analysis, but the data are not available, for this reason we judge there to be a high risk of selective reporting bias.

Other potential sources of bias

We judged all three included studies to be at high risk from other potential sources of bias (Grover 2008; Moran 2001; Moran 2009).

In the Grover study, diabetic patients with normal fasting blood sugars were excluded from the study (Grover 2008). The Grover study was of cross-over design, but we analysed the data as if the study was of parallel design since first-arm data were not available. This approach produces conservative results as it does not take into account within-patient correlation. Also each participant appears in both the treatment and control group, so the two groups are not independent (Elbourne 2002); hence there is an additional risk of bias to the results.

In the 2001 Moran study, the insulin doses were chosen at a pharmacological optimal dose range (0.1 unit/kg/body weight); while repaglinide dosing at 1 mg only represents 25% of the recommended maximal dose (Moran 2001).

In the later Moran study, diabetic patients with fasting hyperglycemia were excluded from the study. Furthermore, after the study was started four repaglinide patients (17%) had doses of medication reduced compared to only two insulin-only patients, which could explain the reason the repaglinide lost statistical significance after the first six months of the study (Moran 2009).

Effects of interventions

In the section below only those comparisons and outcomes for which information or data are available are presented. Data from the cross-over study were not combined with those from the parallel study.

Insulin versus placebo

Two studies presented results for this comparison (Moran 2001; Moran 2009), but we were only able to enter data in the graphs for one of these (Moran 2009). We have presented the results from the earlier study narratively.

Primary outcomes
1. Biochemical measures of glycaemic control

The investigators reported that insulin lispro seemed to have a more beneficial effect than repaglinide on post-meal glucose excursion in CFRD; P < 0.05 when comparing insulin to repaglinide at both two and five hours post-meal (Moran 2001). They reported significant differences between the two drugs in the peak glucose level (P = 0.02), the two-hour glucose area under the curve (AUC) (P = 0.02), and the five-hour glucose AUC (P = 0.01).

The paper also states that at the doses used in the study, neither insulin lispro or repaglinide significantly changed the peak insulin level or the two-hour insulin AUC compared with baseline (Moran 2001).

Peak insulin levels fur AUCs were not measured in the later Moran study (Moran 2009). Specific data were not reported, but the authors commented that postprandial glucose in patients receiving insulin therapy did not achieve statistical significance, and that A1C levels did not significantly change in this group of patients over placebo or when compared to the repaglinide groups (Moran 2009).

2. Pulmonary function

The earlier Moran trial did not report on this outcome (Moran 2001).

In the later trial, Moran reported data for change in FEV1 (% predicted) and FVC (% predicted) at 12 months (Moran 2009). There were no statistical significant differences detected in treatment groups between insulin compared to placebo for CFRD patients; for FEV1 MD 1.20 (95% CI -5.63 to 8.03) (Analysis 1.1) and for FVC MD 0.60 (95% CI -5.67 to 6.87) (Analysis 1.2).

3. Assessment of nutritional status

Only the 2009 Moran trial reported on this outcome and demonstrated a mean change in BMI at 12 months for insulin of 0.39 compared to placebo of -0.02, resulting in a non-significant MD of 0.41 (95% CI -0.23 to 1.05) (Analysis 1.3)

Secondary Outcomes
2. Prevalence of secondary infection complications

The 2001 trial did not report on this outcome (Moran 2001). While the later trial did not specifically report these data, the authors commented that there were no differences in the number of episodes of acute illness during the year of study in either the insulin or placebo arms (Moran 2009).

3. Complications of therapeutic management
a. hypoglycemia (specifically related to oral hypoglycemic and insulin agents)

In the earlier trial, the investigators reported two incidences of hypoglycemia (glucose level 48 to 54 mg/dl, 2.7 to 3.0 mmol/l) after receiving insulin. In all cases catechoamine-related symptoms were present but easily tolerated. These episodes occurred on average four hours after the test meal and resolved spontaneously without glucose administration after 10 to 15 minutes (Moran 2001). When entered in the graphs, these data did not give a significant result, RR 2.00 (95% CI 0.23 to 17.34) (Analysis 1.4).

In the later trial, no serious episode of hypoglycemia was recorded, but in the first three months mild hypoglycemic events were reported in 16% of insulin patients and none receiving placebo (P < 0.04). When entered into the graphs these data did not give a significant result, RR 9.23 (95% CI 0.53 to 159.14) (Analysis 1.4). However, caution is advised on interpreting these results since Moran does not provide information as to whether these events occurred in patients with CFRD or patients with impaired glucose tolerance (IGT). After this time-point, there were no significant differences between groups in the frequency of hypoglycemia, and some patients in the placebo group reported episodes of mild hypoglycemia (Moran 2009).

4. Clinical status
b. HRQoL

Specific data were not included for insulin treated patients from either trial (Moran 2001; Moran 2009); however, the 2009 trial report did state that CFQOL scores did not differ between insulin or placebo groups at baseline or between these groups at the end of the treatment year (Moran 2009).

Repaglinide versus placebo

Only one trial reported on this comparison (Moran 2009).

Primary outcomes
1. Biochemical measures of glycaemic control

It was stated narratively (without actual data) that A1C levels did not significantly change between the repaglinide and placebo groups (Moran 2009).

2. Pulmonary function

Moran reported the change from baseline in FEV1 (% predicted) and FVC (% predicted) at 12 months (Moran 2009). There were no significant differences between the repaglinide or the placebo groups for either outcome: change in FEV1, MD 1.70 (95% CI -5.13 to 8.53) (Analysis 2.1); and change in FVC, MD -1.00 (95% CI -7.40 to 5.40) (Analysis 2.2).

3. Assessment of nutritional status

Moran reported that repaglinide demonstrated a mean difference for BMI change of 0.15 compared to placebo of -0.02, resulting in a MD of 0.17 (95% CI -0.47 to 0.81) (Analysis 2.3).

Secondary Outcomes
2. Prevalence of secondary infection complications

The prevalence of secondary infection was not detailed in the included trial, but the authors did state that there were no differences in the number of episodes of acute illness during the year of study in either the repaglinide or the placebo arm (Moran 2009).

3. Complications of therapeutic management
a. hypoglycemia (specifically related to oral hypoglycemic and insulin agents)

As for the previous comparison, up to three months there were mild hypoglycemic events reported in 23% of repaglinide patients, but no placebo patients reported hypoglycemia (P < 0.04). When entered into the graphs, the result is not significant, RR 12.52 (95% CI 0.74 to 211.20) (Analysis 2.4). However, caution is advised on interpreting these results since Moran does not provide information as to whether these events occurred in patients with CFRD or patients with IGT. After three months, this difference was no longer detected (Moran 2009).

4. Clinical status
b. HRQoL

Specific data were not presented; however, the authors did state that CFQOL scores did not differ between groups at the end of the treatment year (Moran 2009).

c. mortality

Mortality was not specifically reported for this trial, but patients leaving the trial were detailed and none from the repaglinide group nor the placebo group were excluded as a result of mortality (Moran 2009).

Insulin versus repaglinide

Only one trial reported on this comparison (Moran 2009).

Primary outcomes
1. Biochemical measures of glycaemic control

Specific data were not reported, but the authors commented that postprandial glucose in patients receiving insulin compared to repaglinide therapy did not achieve statistical significance, and that A1C levels did not significantly changed between these groups of treated patients (Moran 2009).

2. Pulmonary function

There was no statistically significant difference reported in either FEV1 (% predicted) or FVC (% predicted), MD 0.50 (95% CI -5.60 to 6.60) (Analysis 3.1) and MD -1.60 (95% CI -7.28 to 4.08) (Analysis 3.2) respectively.

3. Assessment of nutritional status

Moran reported that a comparison between insulin and repaglinide demonstrated a non-significant MD in BMI of 0.24 (95% CI -0.34 to 0.82) (Analysis 3.3).

Secondary Outcomes
2. Prevalence of secondary infection complications

The prevalence of secondary infection was not detailed but the authors did state that there were no differences in the number of episodes of acute illness during the year of study in either treatment arm (Moran 2009).

3. Complications of therapeutic management
a. hypoglycemia (specifically related to oral hypoglycemic and insulin agents)

The trial reported mild hypoglycemic events that occurred in 16% of insulin patients and 23% of repaglinide patients at up to three months; after three months, this difference was no longer detected (Moran 2009). Data were analysed in the graphs for the three-month time-point and did not give a statistically significant result, RR 1.38 (95% CI 0.48 to 4.01) (Analysis 3.4). Again, no information was available as to whether these patients had CFRD or IGT.

4. Clinical status
b. HRQoL

Specific data were not presented; however, the authors did state that CFQOL scores did not differ between groups at the end of the treatment year (Moran 2009).

c. mortality

Mortality was not specifically reported for this trial, but patients leaving the study were detailed and none from the repaglinide group nor the insulin group were excluded as a result of mortality (Moran 2009).

NPH insulin versus glargine

Only one trial reported on this comparison; the trial was of cross-over design, but we treated it as a parallel trial since data were not available to analyse them correctly (Grover 2008). There was a washout period of one month which separated either 12 weeks receiving NPH at bedtime or 12 weeks receiving glargine at bedtime.

Primary outcomes
1. Biochemical measures of glycaemic control

Results for neither fasting nor two-hour post-prandial glucose values were significant: fasting glucose, MD 10.00 mg/dl (95% CI -12.86 to 32.86); and two-hour post-prandial, MD 8.00 mg/dl (95% CI -10.07 to 26.07) (Analysis 4.1).

3. Assessment of nutritional status

Grover reported a mean change in weight in the glargine group of 1.2 kg compared to a mean change of 0.2 kg in the NPH group resulting in a non-significant MD -1.00 kg (95% CI -2.39 to 0.39) (Analysis 4.2). Grover also assessed both fat mass and lean mass measured by DEXA scans (Grover 2008). Neither outcome was statistically significant: fat mass, MD -0.30 kg (95% CI -1.41 to 0.81) (Analysis 4.3); and lean mass, MD -0.20 kg (95% CI -0.75 to 0.35) (Analysis 4.4).

Secondary Outcomes
2. Prevalence of secondary infection complications

In the short time course of this study, no infectious complications were reported for either group of patients (Grover 2008).

3. Complications of therapeutic management
a. hypoglycemia (specifically related to oral hypoglycemic and insulin agents)

Grover reported the mean number of minor hypoglycemic episodes per patient, but there was no statistically significant difference between the groups, MD -1.00 (95% CI -3.77 to 1.77) (Analysis 4.5).

4. Clinical status
b. HRQoL

The paper reported that no difference in QoL was observed and that all patients opted to continue treatment with glargine after the trial was completed (Grover 2008).

c. mortality

No mortality was reported for either treatment group; however, there was no reason given for the loss of the patient from the NPH group (Grover 2008).

Discussion

Summary of main results

The prevalence of diabetes in people with CF increases throughout adulthood to rates as high as 50% (Moran 2010), with a seven-fold increase in mortality from this additional complication in the clinical course of disease (Adler 2011). The impact of therapy and the extent to which diabetic control can be achieved is just beginning to be addressed.

We have included three trials in this review (Grover 2008; Moran 2001; Moran 2009); two of which have been added at the 2013 update (Grover 2008; Moran 2009). One of the included trials was a short-term randomized controlled trial comparing insulin to repaglinide (Moran 2001), which has evolved into a multicenter randomized control trial comparing insulin to repaglinide and placebo over a 12-month period (Moran 2009). This later trial identified a slightly more favorable response to insulin and when factoring in the 12-month pre-treatment run-in period the results demonstrated a significant advantage. There were clear differences in the slope of BMI change between the insulin and repaglinide groups during this run-in period. In this review, when factoring out the run-in period which was presented in the original paper, the significance of results vanished and there are no differences between treatment groups (Moran 2009). 

The third included trial compared NPH insulin and glargine over a three-month period (Grover 2008). The only clear conclusion from this comparison that can made at this time, is that while fasting glucose may be diminished by 2 mg/dl for the long-acting insulin glargine, there is no significant therapeutic advantage in treating patients with CFRD and fasting hyperglycemia with glargine insulin compared to protamine Hagedorn insulin (Grover 2008). 

At this point, no conclusions can be made as to the optimal therapy for CF diabetics with normal fasting hyperglycemia; despite recommendations in the current version of the CFRD Clinical Practice Guidelines to treat this subgroup of CF diabetics only at meal times with a short-acting insulin (Moran 2010).

Quality of the evidence

The patient numbers in the studies included in this review were quite small, and the studies were not powered for important clinical outcomes such as FEV1. In the only study where powering was adequate for comparisons of BMI, lead-in bias places some concern as to the validity of the reported outcome (Moran 2009). Unfortunately, there were too many differences between the studies included in the review to allow us to pool data to try to improve on the quality of the currently available evidence.

Agreements and disagreements with other studies or reviews

Some CF centers use oral medications to help control diabetes, yet insulin remains the accepted as the treatment or intervention of choice as reported in the CFRD Clinical Practice Guidelines (Moran 2010) and as evident from the interventions investigated in the limited number of trials that have been evaluated for this review. A cohort study has demonstrated the use of insulin in maintaining a HbA1c below 7% and also a significant reduction in microvascular disease for CFRD (Schwarzenberg 2007); however, studies have yet to significantly demonstrate the long-term impact of insulin therapy on lung function. Outcomes reported in the studies included in this review did not demonstrate a significant impact on lung function between the insulins or oral agents identified from the studies.

The two Moran trials only investigated a subgroup of people with CFRD with normal fasting glucose values (Moran 2001; Moran 2009). Other non-randomized clinical trials of oral hypoglycemic agents have similarly found that these drugs usually only demonstrate a clinical advantage in this subgroup of patients for about the first six months of therapy (Culler 1994; Rosenecker 2001).

Authors' conclusions

Implications for practice

There is currently no significant conclusive evidence that long-acting insulins, short-acting insulins or oral hypoglycemic agents have a distinct advantage over one another in controlling hyperglycemia or clinical outcomes associated with CFRD. A single-center long-term prospective prognostic cohort study is currently the best evidence available to date that demonstrates that insulin therapy for both normal and elevated fasting hypoglycemia improves the control of micro-vascular disease (Schwarzenberg 2007). Outside of this singularly important clinical outcome, the impact on clinical outcomes important to cystic fibrosis patients, namely BMI and pulmonary function measures, have yet to be conclusively demonstrated.

Implications for research

Multicenter, randomized controlled trials are still required in order to best assess the effectiveness of therapeutic options in the control of CFRD. This Cochrane review suggests the potential for achieving tighter glycemic control with long-acting insulin for patients with fasting hyperglycemia and this could likely be demonstrated in future studies with the use of insulin infusion pumps. The use of therapeutic agents which may provide additional impact in CF based on the inflammatory effects of insulin sensitivity and the control of levels of inflammatory mediators by mechanisms such as peroxisome proliferator activated receptor gamma (PPARg) receptor agonists still remains to be demonstrated. The concern with this class of agents is the side effects associated with the glitazones. Cardiac effects associated with the glitazones are less applicable to CF patients, with such a low incidence of coronary artery disease; and osteoporosis concerns may be less problematic in CF due to interleukin inhibitory effect of PPARg agonists on IL-6. Finally, the potential value of metformin in CF patients without hepatic dysfunction demonstrates another potential option in treating CFRD. These additional agents may provide a unique advantage, perhaps as adjuvant therapy to insulin, for a patient population that is highly adversely impacted by inflammatory disease.

Outcomes of greatest interest in future studies comparing long-acting versus short-acting insulins, or insulins to oral agents should be powered such that they can include BMI, FEV1, performance on the six-minute walk test, frequency of hospital admissions and effects on bone metabolism. Furthermore, epidemiology demonstrates increased morbidity and mortality for CF patients with glucose intolerance which prompts RCTs of the use of long-acting insulin in the management of pre-diabetics who are screened as glucose intolerant verses conservative observation.

Data and analyses

Download statistical data

Comparison 1. Insulin versus placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Change in FEV1 (% predicted)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
1.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
2 Change in FVC (% predicted)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
2.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
3 Change in BMI1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
3.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
4 Hypoglycemia2 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
4.1 At 2 months1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
4.2 At 3 months1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 1.1.

Comparison 1 Insulin versus placebo, Outcome 1 Change in FEV1 (% predicted).

Analysis 1.2.

Comparison 1 Insulin versus placebo, Outcome 2 Change in FVC (% predicted).

Analysis 1.3.

Comparison 1 Insulin versus placebo, Outcome 3 Change in BMI.

Analysis 1.4.

Comparison 1 Insulin versus placebo, Outcome 4 Hypoglycemia.

Comparison 2. Repaglinide versus placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Change in FEV1 (% predicted)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
1.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
2 Change in FVC (% predicted)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
2.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
3 Change in BMI1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
3.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
4 Hypoglycemia1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
4.1 At 3 months1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 2.1.

Comparison 2 Repaglinide versus placebo, Outcome 1 Change in FEV1 (% predicted).

Analysis 2.2.

Comparison 2 Repaglinide versus placebo, Outcome 2 Change in FVC (% predicted).

Analysis 2.3.

Comparison 2 Repaglinide versus placebo, Outcome 3 Change in BMI.

Analysis 2.4.

Comparison 2 Repaglinide versus placebo, Outcome 4 Hypoglycemia.

Comparison 3. Insulin versus repaglinide
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Change in FEV1 (% predicted)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
1.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
2 Change in FVC (% predicted)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
2.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
3 Change in BMI1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
3.1 At 12 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
4 Hypoglycemia1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 3.1.

Comparison 3 Insulin versus repaglinide, Outcome 1 Change in FEV1 (% predicted).

Analysis 3.2.

Comparison 3 Insulin versus repaglinide, Outcome 2 Change in FVC (% predicted).

Analysis 3.3.

Comparison 3 Insulin versus repaglinide, Outcome 3 Change in BMI.

Analysis 3.4.

Comparison 3 Insulin versus repaglinide, Outcome 4 Hypoglycemia.

Comparison 4. NPH insulin versus glargine
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Glucose levels (mg/dl)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
1.1 Fasting (at 3 months)1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
1.2 2 hour post-prandial (at 3 months)1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
2 Change in weight (kg)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
2.1 At 3 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
3 Fat mass by DEXA (kg)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
3.1 At 3 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
4 Lean mass by DEXA (kg)1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
4.1 At 3 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
5 Hypoglycemic events per participant1 Mean Difference (IV, Fixed, 95% CI)Totals not selected
5.1 At 3 months1 Mean Difference (IV, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 4.1.

Comparison 4 NPH insulin versus glargine, Outcome 1 Glucose levels (mg/dl).

Analysis 4.2.

Comparison 4 NPH insulin versus glargine, Outcome 2 Change in weight (kg).

Analysis 4.3.

Comparison 4 NPH insulin versus glargine, Outcome 3 Fat mass by DEXA (kg).

Analysis 4.4.

Comparison 4 NPH insulin versus glargine, Outcome 4 Lean mass by DEXA (kg).

Analysis 4.5.

Comparison 4 NPH insulin versus glargine, Outcome 5 Hypoglycemic events per participant.

What's new

DateEventDescription
22 July 2013New search has been performedA search of the Group's Cystic Fibrosis Trials Register identified seven references to four separate studies which were potentially eligible for inclusion in this review; two of these studies (one reference each) have been included (Grover 2008; Moran 2009), while a randomized control trial of a long-acting insulin, was excluded as it only studied cystic fibrosis patients with glucose intolerance and not patients with CFRD (Minicucci 2008); a further study was excluded as it was a study of methods to improve adherence to therapy in children with Type 1 diabetes or cystic fibrosis rather than a comparison of different treatments for cystic fibrosis-related diabetes (Driscoll 2009). Additional searching identified a further study which has been excluded (Hardin 2009).
22 July 2013New citation required and conclusions have changedWhile we are still not able to provide recommendations for treatment based on evidence from randomised controlled trials, we have increased the information available within this review (including a new treatment comparison) and revised our recommendations for future research.

History

Protocol first published: Issue 2, 2004
Review first published: Issue 3, 2005

DateEventDescription
4 February 2009AmendedContact details updated
1 April 2008New search has been performed

A search of the Group's Cystic Fibrosis Trials Register did not identify any new references eligible for inclusion in this review.

A review which was previously excluded has been re-examined and is now included in this update (Moran 2001).

1 April 2008AmendedConverted to new review format.
21 May 2007New search has been performedThe new search identified eight new references to four studies (Borowitz 2005; Konig 2005; Onady 2006; Sulli 2006). None of which was eligible for inclusion in the review and are all listed under 'Excluded studies'.
10 April 2006New search has been performedThe new search identified six new references. However, none of these were eligible for inclusion in the review.
25 May 2005New citation required and conclusions have changedSubstantive amendment

Contributions of authors

Gary Onady conceived the review and drafted the protocol and the full review with additional comments from Adrienne Stolfi.

Gary Onady is responsible for the updates of the review and acts as guarantor of the review.

Declarations of interest

None known.

Differences between protocol and review

Original review

In a post hoc change the comparator group 'oral anti-diabetic agents' were separately listed as follows:

  1. sulphonyureas and related agents

  2. biguanides and related agents

  3. glitazones and related agents

  4. other agents that specifically manage hyperglycaemia.

We originally thought that studies comparing active treatment to placebo would not be undertaken on ethical grounds, however, this proved not to be the case and we have amended our inclusion criteria accordingly.

2013 update

1. The outcome 'Clinical status' has been added as post hoc change at the 2013 update as the authors feel that the six-minute walk test is a more significant measure when powering of studies include low patient numbers. The outcome HRQoL has also been included under 'Clinical status'. Mortality was moved from primary to secondary outcomes as a result of editorial changes regarding the numbers of primary outcome measures allowed.

2. The secondary outcome "Prevalence of secondary infection complications (Pseudomonas aeruginosa/Burkholderia cepacia/Staphylococcus aureus)" has been replaced with the outcome "Rate of pulmonary exacerbations".

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Grover 2008

MethodsRandomized cross-over study.
Single centre.
Duration 12 weeks.
Participants

20 CFRD patients with fasting hyperglycemia only recruited, one dropped out (9 men; 10 women).
Mean (SD) age was 34 (8) years.
Duration of diabetes 9 (5) years; diabetes well-controlled.

Two groups had similar demographic characteristics and gender split.

InterventionsBedtime NPH insulin versus bedtime glargine. Doses were individually established based on an insulin:carbohydrate ratio and a correction scale; however, the ratio and correction scale ranges within the patient population was not reported in the manuscript.  It was noted that the ratio was used to try to maintain target glucose goals between 80-120 mg/dl fasting; 80-150 mg/dl 2 hours after meals.
Outcomes

Total glucose goal of 80-120 mg/dl fasting

Total glucose goal of 80-150 mg/dl 2 hours after a meal

Hemoglobin A1c

Weight

Fat mass (by DEXA)

Lean mass (by DEXA)
QoL
Adverse events

NotesA 1-month insulin adjustment period was included in the study during the cross-over period.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomization sequence details were not provided. A statement was made that groups were similar in demographic characteristics and gender proportion, but a table of baseline comparisons was not provided.
Allocation concealment (selection bias)Unclear riskConcealment of allocation sequence was not discussed in the paper.
Blinding (performance bias and detection bias)
All outcomes
High riskPatients were not blinded. Article states that this was a cross-over study so that once randomized to start either therapy wing (bedtime NPH-Aspart OR bedtime glargine) the patient (and physician) would be aware of the treatment given by a factor of the frequency of dosing. 
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk1 male participant dropped out of the group who received NPH first. However, no reason was given or timing of drop out (which arm).
Selective reporting (reporting bias)High riskAlthough all outcomes stated in the methods section are reported in the results section, data are not available for all outcomes for analysis despite contact with the study authors.
Other biasHigh riskDiabetic patients with normal fasting blood sugars were excluded from the study. This study was of cross-over design but we analysed the data as if the study was of parallel design since first-arm data were not available. This approach produces conservative results as it does not take into account within-patient correlation. Also each participant appears in both the treatment and control group, so the two groups are not independent.

Moran 2001

MethodsRandomised cross-over study.
Single centre.
Duration 3 separate occasions over 1- to 2-month period.
Participants7 (3 male, 4 female) participants with routine oral glucose tolerance test in previous 3 months and CFRD without fasting hypoglycemia.
Mean (SD) age: 24 (5) years.
All were chronically infected with CF pathogens, but no acute illness in previous 3 months.
Interventions3 single-dose treatments over 1 - 2 month period.
No medication vs insulin (0.1 unit/kg) 10 minutes pre-meal vs repaglinide (1 mg) 10 minutes pre-meal.
OutcomesBlood glucose and insulin levels (AUC) at 2 hours and 5 hours post meal, adverse events.
Notes7 healthy age-, sex- and BMI-matched normal control participants in trial, but data for these not presented.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskUnclear, described as treatments given in random order but randomisation method not stated.
Allocation concealment (selection bias)Unclear riskNot discussed.
Blinding (performance bias and detection bias)
All outcomes
High riskParticipants and clinicians would know whether they were given insulin (syringe) or an oral medication or no medication at the time of the test meal. Blinding of outcome assessors is not discussed.
Incomplete outcome data (attrition bias)
All outcomes
High risk

There was no attrition over this brief 5-hour study period.

The paper states outcomes measured at 20 minute intervals, but only presents data for 2 and 5 hour time-points.

Selective reporting (reporting bias)High riskAlthough all outcomes listed in the methods section appear to be reported in the results section of the paper, data are not available for all outcomes for analysis despite contact with the study authors.
Other biasHigh riskThere is potential bias as to optimization of dosing of agents used in this trial. Insulin doses were chosen at a pharmacological optimal dose range (0.1 unit/kg/body weight); while repaglinide dosing at 1 mg only represents 25% of the recommended maximal dose. Otherwise, control of pre-study meals, interventions and measurements were identical in each phase of the study, with the study participants serving as their own controls.

Moran 2009

  1. a

    CF: cystic fibrosis
    CFRD: cystic fibrosis-related diabetes
    DEXA: dual energy X-ray absorption
    FH-: without fasting hyperglycemia
    IGT: impaired glucose tolerance
    NIH: National Institutes of Health
    NPH: Neutral Protamine Hagedorn
    QoL: quality of life
    SD: standard deviation
    vs: versus

MethodsRandomized controlled study. Parallel design.
Multicenter - 14 centres in the USA, Canada and UK).
Duration: 12 months.
Partially blinded (participants blinded to 2 of the 3 treatment arms).
Participants

100 Tanner 5 patients, without acute infection in 2 months prior to the study were enrolled.

Total cohort: 82 participants (61 CFRD (30 male, 31 female; mean (SD) age 28 (9) years); 21 IGT (15 male, 5 female; mean (SD) age 28 (7) years)).

Insulin group: 30 participants (23 CFRD; 7 IGT).

Repaglinide group: 26 participants (22 CFRD; 4 IGT).

Placebo group: 25 participants (16 CFRD; 9 IGT).

A total of 19 dropped out or stopped early (51% female; 49% male) representing 80% of participants completing the study (6 (23%) dropped out from IGT and 13 (18%) from CFRD FH- group) which breaks down to 8 in the insulin group, 6 in the repaglinide group and 5 in the placebo group.

Reasons given were as follows: 5 patients dropped out after becoming positive for fasting hyperglycemia (2 in insulin group; 0 from the repaglinide group; 3 in the placebo group); another 6 discontinued for other medical reasons (1 in the insulin group; 4 in the repaglinide group; 1 in the placebo group); and 8 were non-compliant/quit (5 in the insulin group; 2 in the repaglinide group; 1 in the placebo group).

Interventions

0.5 unit insulin aspart/15 g dietary carbohydrate 3x daily.

2.0 mg repaglinide orally 3x daily (17% had doses reduced due to hypoglycemia).

Placebo orally 3x daily.

All interventions were followed over 1 year.

Outcomes

BMI

Pulmonary function

Body composition by DEXA

CF QOL survey

NIH prognostic score

Hemoglobin A1c
Adverse events

NotesRepaglinide dosing chosen is mid-range and not optimized to maximum dose.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPaper states block randomization using a pseudo-random number generator with stratification by center was used. One third of participants per block were assigned to each treatment arm (page 1784).
Allocation concealment (selection bias)Unclear riskWas not detailed in the study report.
Blinding (performance bias and detection bias)
All outcomes
Low riskBlinding was done for patients receiving oral agent or oral placebo; not for patients receiving insulin.
Incomplete outcome data (attrition bias)
All outcomes
Low riskIndividuals dropping out from the study were detailed. No significant differences in baseline characteristics were evaluated for individuals dropping out, with the exception of the CFQOL measure noting those stopping early had lower eating disturbance scores and social/marginalization scales (P < 0.05).
Selective reporting (reporting bias)High riskAlthough all outcomes were reported and mostly tabulated by diagnosis, all comparing insulin, repaglinide and placebo, data are not available for all outcomes for analysis despite contact with the study authors.
Other biasHigh riskDiabetic patients with fasting hyperglycemia were excluded from the study. Twice as many repaglinide patients (4 or 17%) had doses of medication reduced; compared to 2 on insulin only after the study was started which could explain the reason the repaglinide lost statistical significance after the first 6 months of the study.

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    CF: cystic fibrosis
    CFRD: cystic fibrosis-related diabetes

Borowitz 2005A subset of entire cohort were randomized for pancreatic enzymes, not for diabetic medications. Diabetic outcomes not reported.
Chernoff 2002Open study, participants with diabetes did not have CF and participants with CF did not have diabetes.
Driscoll 2009Study of interventions to promote treatment adherence in children with Type 1 diabetes or CF and not a comparison of treatments for CFRD.
Franzese 2005Not a randomized controlled trial.
Hardin 2009Cohort study, not randomized.
König 2005Not a randomized controlled trial.
Mahroukh 2005Not a randomized controlled trial.
Marshall 2005Not a randomized controlled trial.
Milla 2005Not a randomized controlled trial.
Minicucci 2008A trial of patients with glucose intolerance and not CFRD.
Onady 2006Not a randomized controlled trial.
Peraldo 1998Non-randomized biochemical metabolic study, longitudinal data analysis.
Reali 2006Not a randomized controlled trial, study of a single participant.
Sulli 2007Not a randomized controlled trial, three case studies.
Teeter 2004Study of Type 1 diabetes not CFRD, participants with active lung disease excluded, participants with abnormal pulmonary function at screening excluded.
Ward 1999Not a randomized controlled trial. Treatment of participants with CFRD not detailed, longitudinal data analysis.

Ancillary