We read with great interest the recent manuscript entitled “High Prevalence of the c.74A>C SPINK1 Variant in Miniature and Standard Schnauzers”, by Furrow et al. The authors concluded that they were unable to confirm an association between this SPINK1 variant and clinically detectable pancreatitis. These findings, in part, contradict those of a previous study. However, we feel that the design of the study by Furrow et al had some inherent limitations, which may limit the validity of their conclusions.[1, 2]
One of the most important determinantsfor the success of a genetic association study is the correct assignment of the phenotypes of the individuals enrolled into the study. Incorrect phenotype assignment of even a few individuals (especially when the total number of individuals studied is small) will potentially invalidate the conclusion. In our opinion, in the study by Furrow et al, phenotype assignment (regarding both diseased and healthy control dogs) was suboptimal.
It is of particular concern to us that in many of the dogs assigned to the diseased group (ie, dogs with pancreatitis), pancreatitis was not conclusively confirmed. For example, pancreatitis was diagnosed in 5 dogs solely on the basis of a positive SNAP cPL result. The SNAP cPL test is a semiquantitative test for pancreatitis and the manufacturer of this diagnostic test states that an abnormal SNAP cPLresult indicates that the patient's serum Spec cPL concentration is either in the grey zone (which means that dog may or may not have pancreatitis) or the diagnostic zone (which is diagnostic for pancreatitis). According to the manufacturer's recommendations, all positive SNAP cPL tests should be followed up by the measurement of a quantitative serum Spec cPL concentration to determine whether or not the concentration is in the diagnostic range for pancreatitis.1 Therefore, at least in these 5 dogs, which represent almost 30% of dogs assigned to the pancreatitis group, the diagnosis of pancreatitis is questionable. In our opinion, this is a major shortcoming that has likely affected the results of the study because, as pointed out before, correct phenotype assignment is crucial for a genetic association study. In addition, of the 17 dogs with clinical signs compatible with pancreatitis, 10 dogs underwent ultrasound and in 5 of these dogs pancreatitis was diagnosed based on ultrasonographic findings alone. While the specificity of abdominal ultrasound has been regarded as high, the actual specificity has never been investigated systematically.[3, 4] In a recent study of dogs undergoing laparoscopic pancreatic biopsy for suspicion of pancreatic disease, 23.1% of dogs were thought to have ultrasonographic evidence of pancreatitis, but upon histopathologic evaluation, no evidence for pancreatitis could be identified in any of the dogs.[4, 5] In the same study, there was only a 22% overall agreement between histopathology and ultrasonographic findings.[4, 5]
Of equal importance is the fact that, in the study by Furrow et al, dogs assigned to the healthy control group were not tested for possible pancreatitis. A reasonable control group should have been subjected to the same diagnostic tests used for the diseased population.Recent studies have established that pancreatitis is often not associated with detectable clinical signs.[4, 6] Therefore, it is likely that some of the healthy dogs enrolled in the control group had subclinical pancreatic pathology and would need to be excluded from that group. In the previous genetic association study, several dogs were excluded from the healthy control group because they had an increased serum Spec cPLconcentration, despite the fact that they were free of clinical signs. While one may argue about the clinical significance of an increased serum Spec cPL concentration in clinically normal dogs, an increased serum Spec cPL concentration would raise suspicion of pancreatic pathology of some form and would preclude the assignment of such a dog as a healthy control dog. While this might not be as important for a clinical study, it is of particular importance for a genetic association study. We do recognize that the Spec cPLtest is not 100% sensitive, but it has been shown to be the most sensitive test currently available and the presence of a normal result does carry a high negative predictive value.
Also, when looking at Table 1 of the manuscript by Furrow et al, it is clear that the diseased group had a higher percentage of homozygous dogs than heterozygous dogs and not a single wild type allele. Thus, if the pancreatitis phenotype would have been incorrectly assigned in only a few of the dogs of either group, the P value could easily have changed and reflect the results reported in the previous study. While the authors do mention these shortcomings in the manuscript, we feel that the suboptimal phenotype assignment is an important limitation of this study and may, at least in part, be responsible for the conflicting results with the previous study.
Furrow et al postulate that the SPINK1 variants could possibly affect serum Spec cPL concentrations and thereby explain the findings of the previous study. However, they do not speculate as to any specific physiologic mechanism that could lead to such a finding, nor are we aware of any such a potential physiologic mechanism that could lead to such an effect.