Evolving Mechanistic Views and Emerging Therapeutic Strategies for Cystic Fibrosis–Related Diabetes

John C. Yoon


J Endo Soc. 2017;1(11):1386-1400. 

In This Article

Updates on CFRD Pathogenesis

A. Collateral Damage or Intrinsic β-Cell Defect?

Decreased insulin secretion from the pancreas is the most prominent defect in CFRD. This has historically been attributed to β-cell destruction that occurs in connection with fibrosis and scarring of the exocrine pancreas.[31] According to this "collateral damage hypothesis" or "bystander hypothesis," pancreatic duct obstruction from viscous secretions in cystic fibrosis causes tissue autodigestion by the trapped digestive enzymes, and the progressive destruction and fibrosis of the exocrine pancreas eventually damage the adjacent endocrine cells as a result of spillover of inflammation and compromised blood supply. Exocrine pancreatic insufficiency, or the requirement for pancreatic enzyme supplementation, is a readily discernible consequence of pancreatic destruction and is thus often assessed in clinical studies of CFRD.

Only ~3% of patients with cystic fibrosis are born with exocrine pancreatic insufficiency, whereas most are pancreatic sufficient at birth with relatively mild lesions, such as dilation of the duct and acinar lumen.[35,36] By the end of the first year, about 85% of patients with cystic fibrosis will have developed exocrine pancreatic insufficiency.[14] These individuals typically have two CFTR mutations that result in complete lack of protein function. The other 15% possess at least one copy of the CFTR gene with some residual protein function and will remain pancreatic sufficient or develop exocrine pancreatic insufficiency at an older age.

Patients with exocrine pancreatic insufficiency usually have more severe insulin deficiency, as measured by insulin and glucose response to an OGTT and are much more likely to develop CFRD.[14,30] Studies have consistently shown that even when patients with pancreatic-insufficient cystic fibrosis exhibit normal glucose tolerance on an OGTT, they have lower β-cell secretory capacity and α-cell function than those with pancreatic-sufficient cystic fibrosis or healthy controls.[30,37] Individuals with more severe neonatal exocrine pancreatic disease in utero, as reflected by reduced levels of circulating immunoreactive trypsinogen at birth, have a higher CFRD risk.[38] These observations support the idea that exocrine pancreatic insufficiency is intimately tied to CFRD pathogenesis. Alternatively, it could be a marker of more severe CFTR genotypes that cause diabetes via unrelated mechanisms.

Other data suggest that mechanisms in addition to islet destruction and fibrosis may be at play in CFRD development. The incidence of CFRD in patients with pancreatic sufficiency is much lower than in individuals with pancreatic insufficiency, but it is still higher than the rate of type 2 diabetes in the general population of comparable age and body habitus.[13] Impaired insulin secretion has been demonstrated in patients with pancreatic-sufficient cystic fibrosis, albeit not as severe as in those with the pancreatic-insufficient type.[39] The onset of exocrine pancreatic insufficiency in cystic fibrosis does not immediately herald diabetes, similar to diabetes secondary to chronic pancreatitis. CFRD generally lags behind exocrine insufficiency by one or two decades, reflecting a very slow decline in β-cell mass and functional capacity. A β-cell compensatory response may potentially underlie this phenomenon, as discussed later.[40,41] Autopsy studies have found that the degree of β-cell loss associated with CFRD, which is less than 50%, may not be severe enough to cause overt diabetes on its own.[42,43] Diabetes secondary to pancreatic disorders such as pancreatitis or pancreatic surgery is said to require destruction of 80% of β-cells.[44] Importantly, the extent of islet destruction and fibrosis is not greater in patients with cystic fibrosis with diabetes than in those without,[42] reinforcing the notion that structural damage is most likely not the sole driving factor in the pathogenesis of CFRD.

Autopsy studies have shown that CFRD does correlate with the presence of amyloid deposits in islets; amyloid deposits are absent in islets of patients with cystic fibrosis without diabetes.[45] Amyloid may come from amylin, a polypeptide hormone cosecreted with insulin from β-cells, and whether the islet amyloid contributes to the pathology has not been determined. The islet amyloid accumulation is also frequently seen in type 2 diabetes but generally not in type 1 diabetes. Interestingly, analysis of families with cystic fibrosis has shown that a family history of type 2 diabetes is a risk factor for CFRD, and genome-wide association studies have identified several known susceptibility genes for type 2 diabetes, namely TCF7L2, CDKN2A/B, CDKAL1, and IGF2BP2, as modifier genes for CFRD.[46] These observations suggest that the islet dysfunction in CFRD may share some common mechanisms with that in type 2 diabetes. Indeed, impaired first-phase insulin secretion seen in cystic fibrosis is also found in type 2 diabetes.[31] On the other hand, although insulin resistance is the cardinal hallmark of type 2 diabetes, its role in the development of CFRD is not consistent. Euglycemic hyperinsulinemic clamp studies in nondiabetic subjects with cystic fibrosis have reported conflicting data, finding increased, normal, and decreased insulin sensitivity.[33,47,48] It has been suggested that the insulin resistance found in CFRD may be a secondary consequence of sustained hyperglycemia instead of a primary defect in insulin sensitivity.[42] Acute infections tend to be more consistently associated with increased insulin resistance in cystic fibrosis,[33] presumably because of elevated levels of inflammatory cytokines and stress hormones. In the setting of reduced insulin secretion, changes in insulin resistance may be a major determinant of glucose tolerance.[49]

Aside from β-cell loss from pancreatic destruction and modifier genes shared with type 2 diabetes, some investigators have favored the possibility that a cell-autonomous defect within the β-cell itself directly contributes to the disease process.[50] Intrinsic β-cell defects due to CFTR gene mutations could involve mechanisms such as alterations in cellular membrane potential affecting the insulin secretory apparatus,[51] accumulation of misfolded CFTR protein aggregates producing endoplasmic reticulum stress,[52] or abnormal reduced glutathione transport with increased oxidative stress.[53,54] The CFTR protein is primarily expressed in pancreatic ductal cells, whereas its expression in islet cells has been a subject of debate. Recent work using confocal immunolocalization reported the presence of CFTR protein in human and mouse pancreatic β-cells and α-cells.[55,56] Glucose-stimulated insulin secretion from cultured human and mouse β-cells was significantly inhibited by treatment with CFTR antagonists.[55] In human islets and mouse α-cells, CFTR antagonists increased glucagon secretion in the presence of the cyclic adenosine monophosphate activator forskolin.[56] Depletion of CFTR by short hairpin RNA–mediated gene silencing in a mouse β-cell line reduced glucose-stimulated insulin secretion.[57] However, the expression of CFTR in human endocrine cells and the specificity of the CFTR inhibitor were both questioned in another study.[58]In vivo studies in mice have also produced conflicting results. In one study, the ΔF508 mutant mice displayed attenuated membrane potential and insulin secretion in β-cells isolated from young (12- to 14-week-old) animals, which could be rescued by treatment with the corrector drug lumacaftor.[51] Another study found only a mild β-cell secretory defect in 14-week-old ΔF508 mutant mice, which could be accounted for by a reduction in insulin content, and in older mice found increased insulin resistance and decreased β-cell mass to be the main abnormality, without gross pancreatic pathology.[59] Mice may have limited utility as models of CFRD because they do not spontaneously develop diabetes, although mouse models of cystic fibrosis are more susceptible to streptozotocin-induced diabetes,[60] indicating a baseline abnormality.

In addition to β-cells, α-cells may potentially contribute to dysglycemia in cystic fibrosis. Impaired suppressibility of glucagon after oral glucose has been described in patients with cystic fibrosis, possibly predisposing to early impairment in glucose tolerance.[61] In several studies, patients with cystic fibrosis with exocrine pancreatic insufficiency showed reduced glucagon response to arginine and to insulin-induced hypoglycemia.[30,37] Defective glucagon secretion in cystic fibrosis may increase the risk of hypoglycemia, which is seen even in individuals without CFRD.[22] Other studies have reported normal glucagon response to mixed meals in cystic fibrosis.[62,63] The possibility of an intrinsic α-cell defect has been raised by an observation that treating mouse islets with CFTR inhibitors increases glucagon secretion, perhaps via alteration of the α-cell membrane potential.[56] Another hormonal system that has been implicated in CFRD pathogenesis is the incretin axis. Active glucagon-like peptide-1 (GLP-1) levels in cystic fibrosis are reportedly diminished in some studies.[64] Both GLP-1 and gastric inhibitory polypeptide (GIP-1) responses to a mixed meal were blunted in the first 30 minutes in patients with pancreatic insufficiency compared with patients with pancreatic sufficiency.[37] Whether CFTR is expressed in intestinal enteroendocrine cells and can affect their secretion of GLP-1 or GIP-1 is not known. Treatment of human patients with cystic fibrosis with the CFTR potentiator ivacaftor improved insulin secretion but not incretin secretion,[65] suggesting that CFTR may not directly modulate incretin secretion. Lastly, insulin clearance rate is increased in cystic fibrosis from unclear mechanisms,[33,66] perhaps predisposing to insulin insufficiency.

B. Insights From Ferret and Pig Models

The ferret and pig models mirror the human disease more closely, including age-dependent development of diabetes.[67] Newborn CFTR−/− ferrets have relatively mild disease of the exocrine pancreas and primarily display only acinar duct dilation, but most go on to develop severe inflammation and exocrine pancreatic insufficiency in the first months of life.[68] They serve as a useful model for human infants with cystic fibrosis, most of whom have only mild pancreatic lesions with acinar duct dilation at birth and subsequently undergo pancreatic destruction. The CFTR−/− pigs, on the other hand, develop pancreatic inflammation during late gestation and are all born with exocrine pancreatic insufficiency.[69] The pig is therefore suitable for modeling the later or more severe stages of human pancreatic disease in cystic fibrosis. In terms of the islet pathology, CFTR−/− ferrets exhibit abnormal glucose tolerance and decreased first-phase insulin secretion at birth, before exocrine pancreatic insufficiency.[68]CFTR−/− pigs and ΔF508 pigs also show impaired glucose tolerance and insulin secretion defects at birth and subsequently develop spontaneous hyperglycemia without appreciable loss of islet cell mass.[69] These latter observations imply that structural destruction of the endocrine pancreas may not be required for the development of CFRD, although it is certainly expected to increase the odds of overt diabetes by diminishing the β-cell reserve.

C. How New Human and Animal Data Affect Views on CFRD Pathogenesis

Recent human studies support the presence of early abnormalities in glucose metabolism. Infants and young children aged 3 months to 5 years with cystic fibrosis were found to have abnormal glucose tolerance.[70] Insulin levels were not increased or were only modestly increased relative to controls, suggesting an inability to increase insulin secretion to maintain euglycemia after an oral glucose load. This is reminiscent of the findings in newborn ferret and pig models. In other clinical studies, patients with cystic fibrosis who were given the potentiator drug ivacaftor significantly improved first-phase insulin secretion as well as insulin response to oral glucose load, indicating partial reversibility of the secretion defect.[65,71–73] Such observations could be consistent with, but do not prove, a primary β-cell defect resulting from CFTR mutations because ivacaftor corrects channel defects in other pancreatic cells as well. To exclude influence from non–β-cells, an inducible β-cell–specific CFTR null mouse was generated that exhibited normal β-cell mass, enhanced sensitivity to glucose-stimulated insulin secretion, and evidence of altered endoplasmic reticulum calcium handling.[74] Although the global CFTR knockout mice do not spontaneously develop diabetes with age, they display a higher predisposition to streptozotocin-induced diabetes, and it would be of interest to see whether the β-cell–specific CFTR null mice can replicate that phenotype. Interestingly, the CFTR transcript was detectable only in a minor subpopulation (10% to 20%) of wild-type mouse β-cells by single-cell RNA sequencing.[74] If the CFTR-expressing β-cells possess special pacemaker properties, quantitatively significant β-cell loss may not be necessary for diabetes to develop.

The ferret model resembles early human disease, as noted previously. No CFTR expression was detected in human or ferret endocrine pancreatic cells by single-molecule fluorescent in situ hybridization,[58] conflicting with a recent report of protein detection in human β-cells by confocal immunolocalization.[55] However, islets from newborn CFTR null ferrets still exhibited decreased insulin secretion, as did wild-type islets depleted of CFTR protein by short hairpin RNA, and also showed elevated levels of exocrine ductal markers and markers of stellate cell activation.[58] It was suggested that CFTR may control β-cell function indirectly via a paracrine mechanism involving islet-associated nonendocrine cells, such as duct cells or stellate cells. There may be a role for neuropeptides such as calcitonin gene-related peptide in this context.[75] A paracrine mechanism could account for functional defects in the β-cell without the need to invoke cell autonomous effects or extensive islet destruction. Recent work with the ferrets has also demonstrated the occurrence of a transient glycemic crisis early in life, which is accompanied by loss of β-cell mass and pancreatic inflammation and fibrosis. This is followed by a compensatory response, with a doubling of the residual β-cell mass and enhancement of pancreatic insulin, glucagon, and somatostatin gene expression.[70] The islet protective mechanisms in ferrets are concordant with increased pancreatic expression of the adipogenic transcription factor peroxisome proliferator-activated receptor -γ, which may reflect fatty replacement of pancreatic parenchyma but is hypothesized to have a protective role in conjunction with its anti-inflammatory actions.[70] In the pancreas, adipose tissue stem cells have been reported to promote immunomodulatory and β-cell protective effects.[76]

It is conceivable that human patients with cystic fibrosis may undergo a similar glycemic crisis early in life in the setting of exocrine pancreatic insufficiency, but compensatory mechanisms allow sufficient functional recovery to delay the onset of CFRD until decades later. Autopsy studies have documented prominent nesidioblastosis along with classic fibrocystic changes of the pancreas in nondiabetic patients with cystic fibrosis in the first decade of life.[41]

These new findings collectively point to potential contributions from paracrine mechanisms and β-cell compensation (Figure 1). A role for an intrinsic β-cell defect is supported by experimental studies in human and mouse islets treated with CFTR inhibitors. In ferrets, however, these inhibitors (CFTRinh172) reduced insulin secretion in both wild-type and null islets, suggestive of an off-target effect.[58] It may be informative to perform CFTR expression and inhibition studies using tissues from other CFTR null animal models and patients with CFTR nonsense mutations. The β-cell–specific null mice provide more compelling evidence of an intrinsic abnormality, and a direct connection to CFRD pathogenesis could be more readily established with animal models that spontaneously develop diabetes.

Figure 1.

Possible pathogenetic mechanisms in CFRD. Insulin resistance may also have a role in the setting of infections or glucocorticoid therapy. Modifier genes other than CFTR influence the risk of developing diabetes. ER, endoplasmic reticulum.