Results

A total of 518 patients with cystic fibrosis of median age 26 (range 16–60) attended the Royal Brompton Hospital Cystic Fibrosis Clinic in August 1996 of whom 74 patients were known to be diabetic (prevalence 14%). Between August l996 and May l997, 152 patients attended the Cystic Fibrosis Clinic for annual review. Thirty patients were not eligible to be included in this study including 23 patients who were known to have diabetes. Of the remaining 122 patients (80 male) who were eligible for the study, 91 agreed to participate. Inability to travel to the hospital in the morning due to work commitments or long distance to travel were the usual reasons for the patients’ inability to take part in the study. The demographic details and the results of the OGTTs of the 91 patients studied is shown in table 1. The mean age and body mass index were not significantly different between those who took part in the study and those who were unable to do so. The sensitivities and specificities of individual clinical and biochemical criteria alone in the diagnosis of CFRD, using OGTT as the “gold standard” diagnostic method for CFRD, is shown in table 2.

An abnormal HbA1c was found to have the highest sensitivity (83%; 95% CI 62 to 100) when used alone in the diagnosis of CFRD. By combining the use of different diagnostic criteria, the sensitivity in the diagnosis of CFRD could be increased further (table 3). The highest sensitivity achieved was 92% (95% CI 76 to 100) in patients who had either abnormal RBG (>11.0 mmol/l) and/or abnormal HbA1c (>6.1%) and/or the presence of symptoms of hyperglycaemia or unexplained weight loss. The presence of glycosuria did not increase the sensitivity further. This higher sensitivity was achieved at the expense of a lower specificity of 79% (95% CI 70 to 88) so some patients with one or more of the abnormal criteria cited above may not have CFRD. By selectively performing OGTTs only in this group of patients with one or more of the abnormal criteria cited above, those patients within the group with true CFRD would have been identified and those without CFRD within the group would have been excluded. If such a selective approach was used for the 91 patients studied, only 28 (31%; 95% CI 21 to 41) of the 91 patients would have to undergo an OGTT, but 11 of the 12 patients (92%, 95% CI 76 to 100) with OGTT defined diabetes within the group would have been identified. The twelfth patient had a normal RBG, HbA1c and was asymptomatic and would not have been identified unless an OGTT was performed.

Discussion

Various methods, including OGTT, have been used at different cystic fibrosis centres for the screening and diagnosis of CFRD. At the authors’ unit patients with cystic fibrosis have been screened for diabetes at least annually using RBG and HbA1c for some years. In this study we found that an abnormal RBG (>11.0 mmol/l), when used alone, has a poor sensitivity in the diagnosis of CFRD. In common with the study by Lanng and colleagues,5 we also found that an abnormal FBG and the presence of symptoms of hyperglycaemia when used alone have poor sensitivity in the diagnosis of CFRD. Lanng and colleagues reported that only 16% of their patients with OGTT defined diabetes had abnormal HbA1c. This is in contrast to 83% (95% CI 62 to 100) of our diabetic patients with abnormal HbA1c values. In the Danish study, ion exchange chromatography was used to measure HbA1c and a higher reference range was used (upper limit 6.4%). These differences are insufficient to explain the large differences in the sensitivities between the two studies. HbA1c is a specific type of glycated haemoglobin formed by the reaction of glucose with the N-terminal amino group of the haemoglobin beta chain. It provides an integrated measure of past glucose levels and would be expected to be increased in patients with diabetes.8 We are unable to explain the low percentage of OGTT defined diabetics with abnormal HbA1c levels in the Danish cystic fibrosis patient group. Although the presence of glycosuria has a high specificity in the diagnosis of CFRD, this test is too insensitive to be used as a diagnostic or screening test for the condition. Our study therefore provides further evidence that methods other than OGTT, when used alone, are not sufficiently sensitive or specific in the diagnosis of CFRD. However, performing an OGTT on all patients is very inconvenient for the patients and labour intensive for staff. The need to have an OGTT performed regularly may in some cases decrease patients’ compliance with clinic visits.

In this study we have shown that a selective approach in performing OGTTs has the potential to significantly reduce the total number of patients with cystic fibrosis who would have to undergo OGTTs yet would enable the majority of diabetic patients to be identified. As CFRD occurs almost exclusively in patients with pancreatic exocrine insufficiency,⁹ performing OGTTs in patients who are pancreatic sufficient may not be necessary, thus further reducing the number of OGTTs needed to be performed. The small proportion of patients with OGTT defined diabetes not identified by the selective approach described above would be those with normal RBG, HbA1c, and no symptoms of hyperglycaemia or unexplained weight loss. We speculate that the clinical and metabolic consequences of their diabetes may not be as great as those diabetics with one or more of the abnormal criteria cited above. As most adult patients are reviewed at least three monthly, these patients are likely to be identified at a later date.

Patients with cystic fibrosis already have to undergo a large number of regular investigations. A more selective approach in performing OGTTs in the diagnosis of CFRD described in this study is likely to be welcomed by patients. Further larger studies are warranted to validate this selective approach in the diagnosis of CFRD.