Detection and Management of Early Glucose Abnormalities in Cystic Fibrosis

With advances in technology, it is now possible to detect the emergence of glucose abnormalities in cystic fibrosis with improved sensitivity, and from a very early age. These abnormalities are increasingly recognized as predictors of clinical decline, raising the possibility that early intervention may slow or prevent this deterioration. In this chapter, we will review the available literature on methods of detecting glucose abnormalities in cystic fibrosis (random and fasting glucose, HbA 1c , oral glucose tolerance testing, and continuous glucose monitoring), and detail their advantages and possible limitations in the interpretation of glycemic data. We will also discuss treatment outcomes of early intervention, prior to the diagnosis of diabetes as currently defined.

Taken together, these factors explain why CFRD becomes more common with age. Prevalence is ~1.5% in CF patients aged <10 years, but ~15% in those aged 11-17 and ~50% in those aged ≥18 [8, 16,58]. The American Diabetes Association (ADA) recommends annual screening from age 10, using 2-h OGTT [59]. CFRD can also be diagnosed using clinical status, random blood glucose, fasting plasma  Table 2. Recently, continuous glucose monitoring (CGM) has also been used to investigate glucose abnormalities in CF patients. This method is not yet widely recommended for diagnosis of diabetes, but it is often used to monitor glycemic control or assist insulin dosage [62]. Moreover, CGM often detects even earlier CF-related glucose abnormalities than OGTT, in the form of intermittent postprandial glucose excursions [63]. This chapter compiles research on use of each glucose measurement method in CF patients, with special focus on pre-diabetic patients. The benefits and limitations of each method will be explored to help ascertain when their usage might be appropriate. In the process, we will examine correlations between early glucose abnormalities and clinical decline. Finally, we will review preliminary evidence of improved long-term outcomes with insulin treatment of early glucose abnormalities, supporting their detection and management in routine practice.

Clinical status and/or random blood glucose
The ADA allows diagnosis of CFRD following one random blood glucose measurement ≥11.1 mmol/L, provided that it is combined with polyuria, polydipsia, or hyperglycemic crisis [59]. However, symptomatic hyperglycemia or hyperglycemic crisis is extremely rare in CFRD [4]. In Lanng et al.'s seminal 5-year prospective cohort study of 191 CF patients receiving annual OGTT, only 33% of those diagnosed with CFRD had polyuria or polydipsia [54]. Moreover, in a cross-sectional study of all 60 patients aged ≥10 years at a Brazilian CF center, age at diagnosis was significantly lower for patients diagnosed using OGTT as opposed to clinical criteria (13.5 years vs. 22.3 years), implying much earlier detection of disease [66].
Some centers compensate by accepting unexplained decline in lung function or nutritional status as classical symptoms of hyperglycemia (see Section 3) [67]. In one cross-sectional study of 91 CF patients not known to be diabetic, these modified clinical criteria detected OGTT-diagnosed CFRD with 58% sensitivity [68], which is an improvement over other studies but still suboptimal for a screening test.
Numerous hypotheses have been espoused to explain HbA 1c 's relatively poor correlation with glucose tolerance in CF. These include insufficient duration of transient CF-related post-prandial hyperglycemia, which is often limited to the early phase of insulin secretion; alteration of hemoglobin glycation by hypoxia; iron deficiency, which is a common comorbidity of CF; and increased erythrocyte turnover in the context of chronic inflammation [1,4,5,76,77]. This implies that HbA 1c may vary with degree of inflammation [78], and that trends in HbA 1c may be more useful for predicting deterioration in glucose tolerance. Supporting this, Lanng et al.'s 5-year prospective cohort study found significant differences in median HbA 1c between patients who consistently had NGT (5.2%), patients who varied between NGT and IGT (5.3%), patients who developed CFRD during the study (5.8%), and patients who entered the study with a diagnosis of CFRD (6.5%) [54].
It has also been hypothesized that poor correlation between mean plasma glucose and HbA 1c may be confounded by use of fingerprick tests to measure glucose, since these can easily miss CF-related hyperglycaemic peaks due to their relative infrequency [76]. In two studies of CF and CFRD patients, mean plasma glucose was estimated using 2-7 days of CGM rather than fingerprick self-monitoring, and strongly correlated with HbA 1c (r = 0.86-0.89) [75,79].
Lowering the diagnostic threshold for HbA 1c abnormalities does increase sensitivity to both CFRD and AGT, but the thresholds required to achieve sufficient sensitivity for screening generally have unacceptably low specificity [60]. There is also wide variation in the sensitivities and specificities reported by different studies using the same HbA 1c threshold; this may be due to differences in type of HbA 1c assay [74,86] and timing of the studies relative to the institution's routine OGTT screening [87]. Yung et al., conducting a cross-sectional study of 91 CF patients not known to be diabetic, but also not previously routinely screened, found that HbA 1c ≥6.1% had 83% sensitivity for OGTT-diagnosed CFRD [68]. However, more recent studies with similar designs report only 30-50% sensitivity [39,82,88,89].
Given this uncertainty, the current advice from the ADA is that HbA 1c should not be used to screen for CF-related glucose abnormalities [59]. HbA 1c is still recommended for monitoring glycemic control in CFRD, although normal results must be interpreted with caution [4,78]. It has also been suggested that HbA 1c might be a useful adjunct to OGTT in screening, as its results may fluctuate less and hence, may more accurately predict long-term risk of glucose abnormalities. In a recent 6-year retrospective cohort study of 50 NGT adults with CF followed up with annual OGTT, HbA 1c ≥5.6% had OR 3.49 for development of IGT or CFRD [90].

Fasting glucose
In 2003, the ADA briefly sanctioned fasting plasma glucose as an alternative to OGTT in CFRD screening, because there were insufficient data supporting insulin therapy for CFRD without fasting hyperglycemia [91]. However, subsequent studies have demonstrated similar insulin-induced clinical improvements in patients with and without fasting hyperglycemia [16,92], and treatment of CFRD without fasting hyperglycemia is now standard practice [4]. Only 16-25% of patients diagnosed with CFRD on OGTT have fasting hyperglycemia [8, 54,68,81].
Use of fasting glucose to detect pre-diabetic stages on the glucose tolerance spectrum remains somewhat contentious in CF. Most studies report that fasting plasma glucose does not significantly differ between CF patients with NGT, INDET or IGT [39, 72,93]. The ADA does use fasting glucose to define one pre-diabetic glucose tolerance category, IFG (5.6-6.9 mmol/L), and suggested in 2003 that screening OGTTs could be limited to IFG patients [94]. A prospective cohort study of 1128 CF patients aged 10-64 found that this approach would reduce number of OGTTs by 67%, but miss 17.8% of CFRD and IGT [94]. In a cross-sectional analysis of 73 children with CF, IFG had 100% sensitivity for CFRD, but only 25% sensitivity for IGT [11].
Finally, like HbA 1c , there is debate regarding the utility of IFG as an adjunctive test for predicting long-term risk of CFRD. Frohnert et al. found no significant relationship [95], but Schmid et al. found that IFG generated OR 2.72 for CFRD [96].

Oral glucose tolerance testing
As discussed above, other conventional diagnostic tests have <100% sensitivity for CFRD compared to OGTT. Therefore, OGTT remains the recommended screening test in CF. It is also the only test with standardized definitions of multiple pre-diabetic glucose abnormalities, all demonstrated to predict development of CFRD [96].
Nevertheless, there are several issues with the 2-h OGTT. It may be more inconvenient and resource-intensive than other glucose measurement methods, which is of particular concern in CF because patients and clinics already face a high treatment burden from other aspects of CF care [97]. It also requires patient co-operation, which can be difficult when assessing children [93]. Patients are expected to consume at least 150 g (600 kcal) of carbohydrates for 3 days before an OGTT, then fast for 8 h overnight and be tested early the next morning [59]. They must drink a solution containing a 1.75 g/kg glucose load, preferably within 5 min, then lie or sit quietly for 2 h [64]. In a standard OGTT, venous blood is sampled twice: immediately before ingestion of the load, and at 120 min (BG 120 ). Many CF centers also take hourly or 30-minutely samples to detect post-prandial hyperglycemia that resolves before 2 h [59]. As described earlier, these transient post-prandial glucose excursions are very common in CF, due to selective impairment of early insulin secretion. Our group previously performed OGTT with 30-minutely sampling in 33 children with CF aged 10-19, and found that peak venous insulin concentration was delayed until 90-120 min, producing an early venous glucose peak at 60-90 min [9] (Figure 1).
The inconvenience of OGTT may contribute to poor patient uptake of CFRD screening [98][99][100]. In 2018, the Cystic Fibrosis Foundation Patient Registry reported that the average CF center was screening just 61.3% of adolescents and 32.8% of adults [100]. Rates of utilization of other glucose measurement methods, such as HbA 1c and fasting glucose, were much higher (92.3% for adolescents and 89.6% for adults), suggesting that the main barrier to screening is the OGTT itself [100]. Suggested solutions include shortening the OGTT to 60 or 90 min [83] or replacing it with the 50-g non-fasting 1-h glucose challenge test [89,101], which is currently used to screen for gestational diabetes mellitus in healthy women [101]. These modified OGTT protocols are not standard recommended practice [4].
There are also other issues with the OGTT that likely cannot be resolved by simply shortening it. Its diagnostic thresholds are not specific to CF and may be insensitive to CF-related clinical decline (see Section 3). OGTT results also frequently fluctuate in CF, with a large multicenter prospective cohort study finding a variability coefficient 1.5-1.8 times higher than in the general population [55]. Similarly, in two 4-5 year prospective cohort studies, 18-58% of AGT patients demonstrated overall improvement in glucose tolerance category, while only 14-22% demonstrated deterioration [19,54].
Finally, even with venous sampling at additional timepoints, the peak blood glucose measurements recorded during OGTT may poorly reflect peak blood glucose achieved by CF patients in daily life [4, 60,61]. After all, the OGTT's 1.75 g/kg load contains less glucose than most CF patients' everyday meals [61,98]. This has prompted research into CF-related glucose abnormalities using CGM, a technology that can screen for glucose excursions over a longer interval of everyday life and high calorie CF diet.

Continuous glucose monitoring
Most CGM systems consist of two parts: a sterile sensor, worn subcutaneously for up to 14 days, and a transmitter attached to the sensor that measures interstitial fluid glucose every 30 s, recording an average every 5 min [97] (Figure 2). Some systems do not require a separate sensor, instead measuring interstitial fluid glucose via an electrical current applied across intact skin, but issues have been reported with skin reactions and inaccuracy [102]. Interstitial fluid glucose reflects capillary glucose with a 4-20 min delay [103].
CGM has been validated against OGTT in children with CF of all glucose tolerance categories [104] and non-diabetic adolescents and adults with CF [105]. A subsequent study of this latter group found that they differed significantly from healthy controls in mean CGM glucose (+14.1%) and presence of CGM peaks ≥11.1 mmol/L (+33%), but not in the conventional diagnostic measures of fasting glucose, BG 120 , and HbA 1c [106]. Moreover, 70% of CF patients undertaking simultaneous CGM and OGTT had their CGM peak outside OGTT [106]. This was the beginning of a substantial body of evidence demonstrating the superior sensitivity of CGM to CF-related glucose excursions above OGTT diagnostic thresholds, with numerous studies finding CGM glucose peaks ≥7.8 or 11.1 mmol/L in 71-93% of patients classified as NGT on recent OGTT [14,31,85,98,107,108]. In a 5-year prospective cohort study of 21 adults with CF, 83% had their CGM peak and BG 120 fall in different diagnostic categories, and for 93% the CGM-identified category was worse. Again, this suggests the superior sensitivity of CGM over OGTT [98].
Most of this evidence, particularly in children, is limited by small sample sizes [14, 85,98,107,108] and lack of non-CF controls [14, 85,98,108]. However, it is logical that the increased duration and frequency of glucose monitoring facilitated by CGM, and the opportunity to incorporate the patient's usual diet and physical activity, facilitates more sensitive detection of glucose excursions [109]. CGM is also generally easier and better tolerated than OGTT [78]. While sensors and transmitters are expensive, and staff do require training on their usage, they have become more user-friendly, smaller and cheaper over time [73,110]. The newest devices can be inserted rapidly during a clinic appointment, do not require calibration against fingerpricks, and can be removed by patients or carers without medical supervision [97].
CGM does have one major disadvantage compared to OGTT. The clinical significance of the mild glucose excursions that it detects are still being determined; consequently, there is no standardized system for recognizing and describing clinically relevant CGM findings, and no universally accepted threshold for initiation of treatment [97]. Common variables computed by CGM software include average sensor glucose, maximum glucose, area under the curve of glucose per day (AUC glucose /day), percentage time spent above thresholds (e.g. 7.8 or 11.1 mmol/L), number of excursions ≥11.1 mmol/L, and measures of glycemic variability, such as standard deviation of average sensor glucose [103]. All these parameters have been correlated with HbA 1c in CF patients [75], and many have been correlated with clinical outcomes. However, these studies report heterogeneous findings and rarely include substantial prospective follow-up (see Section 3) [84].
Given all these factors, CGM is not yet widely recommended for CFRD diagnosis or screening [4]. However, it is used in some centers for diagnosis and screening, follow-up of borderline diagnostic tests, and investigation of patients who cannot or refuse to undergo OGTT [31, 111,112]. Like HbA 1c , it may also be useful as an adjunctive test for predicting long-term risk of CF-related glucose abnormalities. In a prospective cohort study of 17 children with CF, all those who had glucose excursions ≥11.1 mmol/L on CGM developed either CFRD or IGT with INDET over a period of 2.5 years, irrespective of their glucose tolerance at baseline [107].

Defining clinically significant sequelae of CFRD: the importance of lung function & nutritional status
CFRD is well-understood to have a differing profile of sequelae as compared to T1D or T2D. Macrovascular disease is uncommon outside of case reports [1,4,5,113], and although screening for microvascular disease should be routinely undertaken [59], microvascular complications are uncommon until at least 5-10 years of CFRD with fasting hyperglycemia [57,114,115]. Therefore they are substantially predated by declines in lung function [6-21, [116][117][118] and nutritional status [7, 9-12, 14, 117], both of which are significant predictors of early mortality in CF [10,11,16,18,25,56,119]. Four large cohort studies also report higher annual frequency in diabetic vs. non-diabetic CF patients of pulmonary exacerbations requiring intravenous antibiotics or hospitalization [10,21,39,120], and it was recently demonstrated that diabetic CF patients have reduced recovery of baseline forced expiratory volume in 1 sec as a percentage of predicted (FEV 1 %) following pulmonary exacerbations [116].

Decline in clinical status prior to diagnosis of CFRD
Numerous cohort and case-control studies examining the 1-5 years before CFRD diagnosis report decline in lung function [19,35,38,92,[143][144][145][146] and nutritional status [19,35,38,92,143,144] in pre-diabetic patients, or significantly reduced values compared to non-diabetic CF controls [12,17]. This suggests that prediabetic glucose abnormalities are clinically significant. Two case-control studies focusing specifically on pediatric populations also report that pre-diabetic children with CF have significantly lower height and weight velocities than non-diabetic CF controls [145,146], with one study demonstrating differences up to 11 years before CFRD diagnosis [146]. These differing velocities produce steadily-widening gaps in height-for-age and weight-for-age, reaching statistically-significant sizes after CFRD diagnosis, usually around ages 15-19 [18,146]. Importantly, this growing disparity seems to occur even if aggressive insulin therapy is commenced at diagnosis [144], and although it may narrow with prolonged therapy, it may not fully correct [18, 144,147]. Therefore, optimizing clinical outcomes in CFRD may require treatment of pre-diabetic abnormalities, highlighting the importance of glucose measurement systems that can sensitively predict clinical decline.

Clinically significant pre-diabetic markers detectable using OGTT
Traditional OGTT diagnostic thresholds are not specific to CF -in fact, they were originally designed to predict T2D-associated microvascular disease in Pima Native Americans [148]. This may explain their apparent insensitivity to CF clinical outcomes. A few studies do report poorer lung function or nutritional status in IGT vs. NGT CF patients [37,72], and several more identify IGT as a significant risk factor for substantial decline in FEV 1 % over 4-5 years [19,149]. However, most studies attempting to correlate IGT with contemporary lung function and nutritional status find no significant relationship [19,33,34,39,53,[70][71][72][73][150][151][152].
A more successful non-conventional OGTT parameter is the additional glucose tolerance category of INDET, defined as blood glucose ≥11.1 mmol/L at an OGTT midpoint -most commonly 60 min (BG 60 ) -as opposed to 0 or 120 min [4]. BG 60 has been shown to inversely correlate with BMI in children with CF, and correlates with FEV 1 % and forced vital capacity as a percentage of predicted (FVC%) in both children [7] and adults [150]. In a subsequent study, INDET patients had mean FEV 1 % comparable to CFRD patients, representing a significant reduction compared to NGT and IGT patients [71]. INDET has also been confirmed to predict development of CFRD (OR 2.81 over ~3.5 years) [93,96].
Other OGTT parameters shown to predict FEV 1 % in non-diabetic CF patients include higher peak glucose (BG max ) [9, 33, 72, 153], higher AUC glucose [124,153], and reduced insulin secretion [34,35,72,124]. Finally, a few studies have correlated FEV 1 % with trajectories of deterioration in glucose tolerance [41,154]. One prospective cohort study recruited 152 non-diabetic CF patients, and stratified them according to whether their glucose tolerance on OGTT improved, deteriorated or remained stable over 2 years [41]. While all patients experienced a decline in FEV 1 %, the extent of decline only reached statistical significance in patients of stable or deteriorating glucose tolerance, and those of deteriorating glucose tolerance also had a much larger drop than those of stable glucose tolerance (−6.1% vs. −1.6%) [41].
It is rarer for studies to report correlations between OGTT parameters and nutritional status [33-35, 41, 71, 72, 154], possibly because intensive dietician management of CF mitigates nutritional decline [133,154]. Nevertheless, one seminal prospective cohort study inversely correlated age-adjusted height and BMI with AUC glucose [8], and a recent cross-sectional study found that lower-thanmedian insulin secretion at 60 min is independently associated with worse BMI [150]. In children, BMI (calculated as weight in kg divided by the square of height in meters) may be a less sensitive measure of nutritional status than weight-for-age, as poor linear growth may mask decline [146]. Nevertheless, Wooldridge et al. report a direct correlation between AUC insulin and BMI z-score in 146 NGT children with CF aged 5-20 [123], and our group has found that AUC glucose inversely correlates with age-adjusted weight, height and BMI in children aged ≤10 years [153]. Furthermore, in an earlier cohort study of 33 children aged 10-19, we found that higher BG max was associated with decline in weight z-score, FEV 1 % and FVC% over the past 12 months, and BG max ≥8.2 mmol/L had 87% sensitivity and 70% specificity for a clinically significant decline in weight z-score [9]. By contrast, BG 120 was no better than chance at detecting decline in weight z-score, and the conventional diagnostic threshold of 11.1 mmol/L had only 10% sensitivity [9]. These findings led us to propose an alternative system for classifying CF-related glucose abnormalities on OGTT, the Cystic Fibrosis Insulin Deficiency (CFID) stages (Table 3) [9].

Clinically significant pre-diabetic markers detectable using CGM
Six main studies have explored the clinical significance of CGM-based measures of CF-related early glucose abnormalities [9, 98,111,152,155,156]. Their results are compelling but heterogeneous. Taylor-Cousar et al. conducted a 5-year prospective cohort study of 17 originally non-diabetic CF patients, 7 of whom developed CFRD during observation [98]. In this subgroup, there was significant inverse correlation between peak glucose and BMI, and a trend towards correlation with FEV 1 % [98]. Leclercq et al. also examined peak glucose, stratifying 38 NGT CF patients according to whether they had any peaks ≥11.1 mmol/L during 72-h CGM [155]. In the 'yes' group, there was significantly lower FEV 1 % and FVC%, and increased risk of colonization with P. aeruginosa [155].
In the aforementioned study undertaken by our research group in 33 children with CF aged 10-19, we also showed that percentage time ≥7.8 mmol/L on CGM predicted 12-month rate of decline in weight z-score, FVC%, and FEV 1 %. Similarly, on receiver operator characteristic (ROC) analysis, ≥4.5% time at ≥7.8 mmol/L on CGM was a sensitive and specific predictor of clinically significant decline in weight z-score and FVC% [9]. Frost et al. subsequently used these parameters to interpret the CGM results of 59 adults being investigated for CF-related glucose abnormalities [112]. They found that percentage time ≥7.8 mmol/L on CGM correlated with baseline FEV 1 % and 12-month rate of decline [112].
In Chan et al.'s study of 88 children with CF aged 10-18, 12-month decline in FEV 1 % and FVC% was predicted by multiple other CGM parameters: peak glucose, number of daily glucose excursions >11.1 mmol/L, mean amplitude of glycemic  excursions, and standard deviation [152]. Brugha et al. investigated another glycemic variability measure, glucose interquartile ranges, in a 7-year retrospective cohort study [111]. On ROC analysis, ranges >1.95 mmol/L predicted CFRD with 60% sensitivity and 98% specificity, but did not correlate with BMI or FEV 1 % [111]. Finally, our group recently conducted a cross-sectional study of 18 children with CF aged ≤5 years [156]. Even in this very young group, history of P. aeruginosa was predicted by mean glucose and percentage time at ≥7.8 mmol/L, and levels of inflammatory markers in bronchoalveolar lavage fluid were predicted by peak glucose, mean glucose, percentage time at ≥7.8 mmol/L, and standard deviation [156].

HbA 1c and alternative glycated proteins
Three studies report a weak inverse correlation between HbA 1c and lung function in non-diabetic CF patients (r = −0.25-0.3) [72,73,88], and one of these also found a direct correlation with number of infective pulmonary exacerbations per year [73]. In two more studies, HbA 1c ≥ 5.5-5.8% predicted poorer FVC% [74] or FEV 1 % [82]. Therefore HbA 1c , despite its insensitivity to CF-related glucose abnormalities, may be a useful harbinger of clinical decline when elevated.
Several studies have also investigated fructosamine, glycated albumin, and 1,5-anhydroglucitol as alternatives to HbA 1c in CF. These biomarkers are not dependent on the lifespan of erythrocytes, and have been shown to correlate with mean plasma glucose in CF as estimated using CGM [75]. However, evidence of their ability to predict glucose abnormalities and clinical decline in CF is currently mixed [11,74,157]. In one study, fractional serum fructosamine (FSF) ≥3.70 μmol/g predicted IGT and CFRD with 100% sensitivity and 67% specificity, and patients with elevated FSF also had significantly lower median FEV 1 % (47% vs. 90%) [157].

Fasting glucose
Early evidence suggests that fasting glucose, including IFG, does not correlate with clinical status in CF [53,95]. In one case-control study, IFG actually predicted better lung function than normal fasting glucose in some patient subgroups, particularly children with simultaneous IGT [95]. It was hypothesized that IFG may represent a physiological adaptation to CF, with hepatic glucose production upregulated to meet increased baseline metabolic requirements [95].

Detection protocols for early glucose abnormalities and CFRD at the Sydney Children's Hospital, Randwick
Our institute, the Sydney Children's Hospital, provides one example of integrating multiple glucose measurement methods into routine practice. Children with CF are screened annually for glucose abnormalities from age 10, using OGTT with 30-minutely sampling. CGM is used to follow up borderline OGTTs, or to investigate children with clinically-suspected glucose abnormalities who have normal OGTTs or are unable to undergo OGTT. CGM excursions ≥11.1 mmol/L over 72 h of monitoring are considered severe abnormalities that warrant further investigation for possible insulin therapy. Moreover, some pre-diabetic children on OGTT are randomized to insulin therapy via the CF-IDEA trial (ClinicalTrials. gov Identifier NCT01100892, see Section 5).

Management of early glucose abnormalities in CF
Ultimately, the most clinically relevant measures of CF-related early glucose abnormalities are those that alter patient management. Therefore the long-term effects of actively treating early abnormalities is an important research question. Most studies have focused on insulin therapy, as insulin is currently the only recommended pharmacotherapy for CFRD (in part because of its anabolic effects) [59]. Emerging research has also explored oral anti-hypoglycemics [158], incretin modifiers [159], and CFTR modulators [160,161].
It is already known that earlier diagnosis and treatment of CFRD, via OGTT screening programs, improves life expectancy and resolves historical sex differences in clinical outcomes (females with CFRD previously did worse than males) [16,24]. Seven studies were identified trialing insulin therapy for CF patients who were pre-diabetic on OGTT [92,122,143,[162][163][164]. Five report statistically-significant improvements in lung function [122,163,165], nutritional status [122,143,164,165], or rate of decline in either variable [163,164], either intra-individually or relative to untreated controls. Moreover, five out of six studies assessing tolerability found no significantly-increased incidence of symptomatic hypoglycemia [92,122,143,162,164,165]. Finally, one additional study has assessed the efficacy of insulin therapy initiated based on CGM, via retrospective analysis of all non-diabetic adults at a British CF center who had a CGM ordered between 2013 and 2016 [112]. Insulin was initiated if patients spent >4.5% time at >7.8 mmol/L on CGM, and if they recorded no clear triggers for these glucose excursions in a contemporary food diary. Patients treated with insulin demonstrated statisticallysignificant improvements in FEV 1 % and weight within 3 months of treatment, and maintained an improvement in weight and annual rate of lung function decline at 12 months [112].
All this suggests that treatment of CF-related AGT may be beneficial. However, results are difficult to generalize, due to heterogeneity in studies' inclusions criteria, types of controls, and insulin regimens [166]. Studies are also limited by small sample sizes [92,112,122,143,[162][163][164][165], short durations [92,112,122,143,162,165], and mixed analysis of pre-diabetic and diabetic patients [92,122], highlighting the need for large long-term randomized control trials. One such trial, CF-IDEA (ClinicalTrials.gov Identifier NCT01100892), is nearing completion. To date, CF-IDEA has recruited 86 participants aged ≥5 years at 5 participating sites, all non-diabetic on OGTT with BG max 8.2 mmol/L to <11.1 mmol/L (CFID1) or ≥ 11.1 mmol/L (CFID2). Participants are randomized to observation only or to a once-daily insulin detemir (Levemir) for 12 months, with starting dose 0.1 units/kg/day, blood glucose self-monitoring intensively for 10 days and twice daily thereafter, and a blood glucose target range of 4-8 mmol/L. The main outcome factors are change in weight SDS, change in lung function, and frequency of hospitalization.

Conclusions
As patients with CF live longer, CFRD becomes an increasingly prevalent serious co-morbidity, associated with significant decline in lung function and nutritional status. Evidence suggests that this decline may begin years earlier, in the pre-diabetic phase. Currently, OGTT is the most sensitive licensed diagnostic tool for identifying pre-diabetic CF-related glucose abnormalities, but its utility is limited by inconvenience, high variability of results, and insensitivity of traditional diagnostic categories to CF-related glucose excursions and clinical decline. from Novo Nordisk, Medtronic, and Abbott Diagnostics. BP has been awarded a fellowship from the Thoracic Society of Australia and New Zealand and Vertex, and a postgraduate scholarship from the National Health and Medical Research Council of Australia.
© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.