Evaluation of Hypertriglyceridemia as a Mediator Between Endocrine Diseases and Pancreatitis in Dogs
ABSTRACT
The role of hypertriglyceridemia (HTG) secondary to endocrine diseases in the occurrence of pancreatitis in dogs has not been fully investigated. The objective of the present study was to evaluate HTG as a mediator between endocrine diseases and pancreatitis in dogs. The study design was a retrospective case-control study. Medical records of dogs newly diagnosed with acutely presenting pancreatitis between 2012 and 2014 were reviewed for the presence or absence of hyperadrenocorticism (HAC), diabetes mellitus (DM), and hypothyroidism. A matched case-control analysis was performed, and the association between endocrine diseases and pancreatitis was evaluated using multiple logistic regression analysis. In dogs with pancreatitis, the odds of HAC (P < .001) and DM (P < .001) were 4.5 and 12.4 times that of dogs without pancreatitis, respectively. HTG significantly mediated the association between DM and pancreatitis but not between HAC and pancreatitis. Additional studies will be necessary to confirm these findings and to further elucidate the associations between endocrine diseases and pancreatitis.
Introduction
Pancreatitis refers to an inflammatory process in the pancreas, with variable involvement of peripancreatic tissues or remote organ systems.1,2 Many risk factors associated with pancreatitis have been identified or are suspected in humans and dogs.2−4 Hypertriglyceridemia (HTG) has been accepted as a possible risk factor for naturally occurring pancreatitis in dogs, although dogs with experimentally induced pancreatitis did not develop HTG, suggesting that HTG might be a cause of pancreatitis rather than a consequence.5–8 However, the previous studies evaluated primary HTG, and there is a lack of data concerning the relationship between secondary HTG and pancreatitis in dogs. Although a definitive cause-and-effect association has not been established, the final common pathway in all cases is the abnormal early activation of trypsinogen and other pancreatic proteases.1,4,8 Endocrine diseases such as diabetes mellitus (DM), hyperadrenocorticism (HAC), and hypothyroidism have been suggested to be associated with an increased risk of developing pancreatitis, or an increased risk of pancreatitis becoming fatal.2,4,8 However, it has only been reported to date that endocrine diseases either increase the risk of pancreatitis or increase the risk of it being fatal, indicating that dogs are more likely to die from pancreatitis when they have such diseases.8 Therefore, to the authors’ knowledge, no previous report has definitively shown that endocrine diseases can increase the risk of acutely presenting pancreatitis in dogs. In addition, there has been much debate regarding whether endocrine diseases can directly affect the development of pancreatitis.3,8
Several studies of lipid metabolism disorders in dogs with pancreatitis have been published, and associations between endocrine diseases, such as HAC, DM, hypothyroidism, and HTG, are also described.5–7,9,10 However, to the authors’ knowledge, it is unclear if endocrine diseases cause pancreatitis via secondary HTG or if they affect the risk of pancreatitis independently of HTG. Therefore, we hypothesized that untreated endocrine diseases might be a risk factor for pancreatitis in dogs, and that HTG resulting from endocrine diseases predisposes dogs to pancreatitis. The aims of this retrospective study were to examine the prevalence of endocrine disorders (HAC, DM, and hypothyroidism) and to evaluate HTG as a mediator between endocrine diseases and pancreatitis in dogs.
Materials and Methods
Case Selection
This study was a retrospective case-control study. The medical records of dogs who were presented to the Veterinary Medical Center between September 2012 and July 2014 were reviewed. Among the 186 dogs diagnosed with acutely presenting pancreatitis, 82 dogs were excluded from the study for one of the following reasons: incomplete or missing medical records; presumptive diagnosis of endocrine disease based on history and clinicopathological and radiographic findings, but not those based on the results of further endocrine tests; the presence of concurrent renal or hepatic diseases; indeterminate duration of fasting; or dogs receiving drugs to directly manage their endocrine disease. These drugs were insulin therapy for DM; any medication including mitotane, trilostane, or ketoconazole for HAC; and levothyroxine for hypothyroidism. Consequently, the pancreatitis group consisted of 104 dogs, who were divided into two groups according to the presence of HTG, and additional statistical analyses were conducted in the two subgroups.
The control group also consisted of dogs admitted between September 2012 and July 2014. Only dogs having negative results in the SNAP canine pancreas-specific lipase (cPL)a and Spec cPLa tests were included. Dogs with suspected pancreatitis, but who had not undergone further diagnostic testing, and those with a history of pancreatitis were excluded from the control group. Furthermore, dogs with gastrointestinal signs, including vomiting, anorexia, diarrhea, and/or abdominal pain, who had concurrent diseases that might influence lipid metabolism (e.g., renal and hepatic diseases) and well-controlled endocrine diseases were also excluded. Each control was age- (within ±3 yr) and breed-matched to a dog with pancreatitis. When a dog with pancreatitis was of an uncommon or mixed breed, a size-matched control (within ±3 kg) was selected. The sample size calculation was as follows: A total of 104 dogs with pancreatitis were included, each of whom were matched with two controls. Assuming that the frequency of endocrine diseases was 34% among dogs with pancreatitis based on a previous study, an odds ratio (OR) of 4.0 would be detectable with α = 0.05 and power = 0.80.8
Review of Medical Records
The following data was extracted from the medical records of pancreatitis cases and controls: age; breed; sex; body weight; the presence of DM, HAC, and hypothyroidism; and the presence of HTG.
Diagnoses
A diagnosis of pancreatitis was established only if all the abnormal findings were compatible with acute onset of the disease, that is, abnormal findings consistent with pancreatitis on ultrasound, a positive SNAP cPL test, and increased Spec cPL concentrations, as described elsewhere.12–14 Dogs exhibiting only some of these diagnostic criteria were not included in the study. Clinical evidence of acutely presenting pancreatitis was confirmed if acute-onset (<2 days) vomiting, anorexia, and/or abdominal pain were present during physical examination on admission, or when history was established.13 Ultrasonographic findings suggestive of pancreatitis, including hypo/hyperechoic lesions or mixed patterns, were recognized in the potentially enlarged and irregularly shaped pancreata. In addition, changes typically recognized as being secondary to pancreatitis, such as hyperechoic mesentery, localized free abdominal fluid, thickened duodenal or gastric wall, duodenal spasm, irritated appearance of the adjacent intestines, and dilated common bile duct, were also considered to be evidence of pancreatitis by ultrasonographyb.13 The results of SNAP cPL tests were interpreted as abnormal only if the color of the sample spot was more intense than that of the reference spot. In the Spec cPL assays, concentrations >400 μg/L were considered to be consistent with pancreatitis.15
DM was diagnosed using the following criteria: history of polyuria and polydipsia, weight loss despite a good appetite, persistent glycosuria, increased serum fructosamine concentrationa, and a persistent fasting blood glucose concentration >250 mg/dL.16 The presence of DM was considered to be established when a historical diagnosis was recorded or if DM was diagnosed during hospitalization with pancreatitis.
HAC was diagnosed as described elsewhere.17,18 A tentative diagnosis of HAC was based on the history, abnormal findings during physical examination, hematology, biochemistry, and urinalysis at admission. All dogs with suspected HAC were also evaluated by abdominal ultrasonographyb. Biochemical measurements were made using an automated analyzerc, and electrolytes were assayed using an electrolyte analyzerd. All of the dogs diagnosed with HAC showed at least four of the following abnormal findings: high serum alkaline phosphatase activity, high serum alanine aminotransferase, hypercholesterolemia, HTG, hyperglycemia, and urine-specific gravity <1.020. All dogs with suspected HAC underwent an adrenocorticotropic hormone stimulation test and a low-dose dexamethasone suppression test. Serum cortisol concentrations were analyzed using a chemiluminescent immunoassay-based analyzere. To exclude the possibility of false-positive results caused by pancreatitis, a diagnosis of HAC was considered to have been established when a historical diagnosis was recorded, when HAC was suspected during hospitalization for pancreatitis, or after the resolution of pancreatitis, and a definitive diagnosis was made during follow-up investigations. If there was any suspicion that a dog in either group might have had HAC at presentation, but the initial results were not diagnostic, the listed diagnostic tests were conducted again at a 3–6 mo follow-up appointment.
Hypothyroidism was diagnosed based on history, clinical signs, and thyroid function tests.19–21 Total thyroxine (T4) was measured in all of the included dogs. Serum total T4 and thyroid-stimulating hormone (TSH) concentrations were measured using a chemiluminescent immunoassay-based analyzere. Serum TSH was measured in dogs with a total T4 below the reference range (1.0–3.5 μg/dL), and dogs with low total T4 and TSH above the reference range (0.01–0.6 ng/mL) were diagnosed as hypothyroid. However, dogs with a total T4 below the reference range and a normal TSH concentration underwent a TSH stimulation test. If a dog had a low total T4 concentration with a normal or low TSH concentration but had not undergone a TSH stimulation test, the dog was excluded.
The results of the TSH stimulation tests were interpreted according to criteria established in previously published studies.19,20 Briefly, a post-TSH total T4 concentration of >2.5 μg/dL exceeding at least 1.5 times the basal total T4 concentration was considered to be an euthyroid test result. TSH stimulation test results with a post-TSH total T4 concentration between 1.6 and 2.5 μg/dL and a post-TSH total T4 concentration of >1.5 times the basal total T4 in dogs without clinical signs of hypothyroidism (lethargy, inactivity, weight gain, alopecia, seborrheic dermatitis, muscle weakness, anemia, hyperlipidemia, and/or bradycardia, etc.) were also considered to be normal.21 Dogs with a post-TSH total T4 concentration of <1.6 μg/dL and <1.5 times the basal total T4 were diagnosed with hypothyroidism. Dogs with post-TSH total T4 concentrations between 1.6 and 2.5 μg/dL or post-TSH total T4 concentrations of ≥2.5 μg/dL, but with less than a 1.5-fold increase in the basal T4 concentration, were considered to have intermediate stimulation.19 However, because the dogs only had signs suggestive of hypothyroidism, the dogs were tested again at a 3–6 mo follow-up appointment. To exclude the possibility of false-positive results caused by pancreatitis, dogs were determined to have hypothyroidism if a diagnosis was made during hospitalization with pancreatitis using TSH stimulation test, to differentiate euthyroid sick syndrome, or alongside the follow-up investigations for pancreatitis.19 If a dog in either group was suspected to have hypothyroidism at presentation but the initial results were not diagnostic, the listed diagnostic tests were conducted again at the 3–6 mo follow-up appointment.
HTG was defined as serum triglyceride concentrations persistently above the reference interval (21–116 mg/dL). Blood samples for HTG testing were obtained after at least 12 hr of fasting. Dogs were considered to have persistent HTG when HTG was observed more than twice during the follow-up conducted at ∼2 and 6 wk after the resolution of pancreatitis. Therefore, persistent HTG was considered if HTG was present at least 6 wk after the resolution of pancreatitis.
Statistical Analyses
Data was analyzed using a commercially available statistical programf,g. After conducting a normality test (Kolmogorov-Smirnov test), continuous variables (age and body weight) were expressed as median (range). Categorical variables were expressed as frequencies in the pancreatitis and control groups. Age and body weight were analyzed using the Mann-Whitney U test. Pearson’s χ2 test was conducted to examine the differences in the sex ratio and neutering status as well as the frequencies of endocrine diseases and HTG between the pancreatitis and control groups. After univariate analyses, variables with P < .20 were included in subsequent multivariate analyses. OR and 95% confidence intervals (CIs) for the association of endocrine diseases with pancreatitis were estimated using multiple logistic regression analyses.
Pearson’s χ2 tests or Fisher exact tests were used to compare the proportions of dogs with endocrine diseases (DM, HAC, and hypothyroidism) and HTG between the pancreatitis and control groups.
Additionally, to test whether the association between endocrine diseases and pancreatitis was mediated by HTG, mediation was evaluated using the Baron-Kenny approach.22–24 The Baron-Kenny approach to establish moderating and mediating factors is based on the model
where ε = the variance of the error term, independent of both X and M; and XM = the interaction term that is created by X Χ M. In this model, the Baron-Kenny approach assumes that the interaction between X and M is zero in the population for mediation (β3 = 0) and, thus, does not include the interaction in the model. However, if the result shows that the interaction term is statistically significant (β3 > 0), this indicates that M is a moderator, but not a mediator, of the relationship. Conceptually, a variable M is a mediator of the relationship between X and Y if M helps explain how or why X is related to Y. In contrast, a variable M is a moderator of the relationship between a target variable (X) and an outcome (Y) in a particular population if M explains under what conditions X is related to Y.
The mediation proportions were calculated by estimating the percentage change in the regression coefficients when an intermediate variable, HTG, was included in the model.24 In the present study, the first step (pathway c) of the Baron-Kenny approach was to test the hypothesis that dogs with the endocrine diseases (DM and HAC) have a higher risk of pancreatitis than dogs without endocrine diseases, which used the same analysis as the multiple logistic regression analyses described above. The second step (pathway a) was to evaluate the association between the endocrine diseases (DM and HAC) and HTG. The third step (pathway b) was to evaluate whether HTG was a mediator of pancreatitis when the effect of the endocrine diseases (DM and HAC) was controlled for. The final step (pathway c’) was to assess whether the endocrine diseases were significantly associated with pancreatitis after accounting for the effect of HTG. If the effect of the endocrine diseases remained significant when HTG was controlled for, the mediation was considered partial. When controlling for the effect of HTG rendering the effect of endocrine diseases nonsignificant, mediation was considered complete. Furthermore, if the interaction term is not statistically significant, the HTG can be regarded as a mediator, whereas if it is statistically significant, the HTG can be regarded as a moderator. P < .05 was considered statistically significant.
Results
The medical records of 186 dogs diagnosed with acutely presenting pancreatitis between September 2012 and July 2014 were examined, and 82 dogs were excluded. Consequently, 104 dogs with acutely presenting pancreatitis were included in the present study. The basic characteristics of these dogs (pancreatitis group) are presented in Table 1.
As matched control group, 208 dogs were included. Further characteristics and the diagnoses made in dogs in the pancreatitis and control groups enrolled during the same period are presented in Table 2. Age (P = .68) and body weight (P = .39) were not statistically different between the pancreatitis and control groups. There were no differences in sex ratio and neutering status between the two groups (P = .69). Twenty dogs in the pancreatitis group (19.2%) and 12 dogs in the control group (5.8%) were diagnosed with HAC. Twenty-six dogs in the pancreatitis group (25.0%) and 6 dogs in the control group (2.9%) were diagnosed with DM. One dog in the pancreatitis group (1.0%) and 4 dogs (1.9%) in the control group were diagnosed with hypothyroidism. Thirty-four dogs and 33 dogs were hypertriglyceridemic in the pancreatitis (32.7%) and control (15.9%) groups, respectively.
There were differences in the frequency of DM (P < .001; 95% CI 4.5–28.3) and HAC (P < .001; 95% CI 1.8–8.3) between the pancreatitis and control groups (Table 2). However, there was no difference in the frequency of hypothyroidism between the two groups (P = .67). As expected, there was a significant difference in the frequency of HTG (P < .001; 95% CI 1.5–4.5). Subsequently, to examine an association between endocrine diseases and pancreatitis, a multiple logistic regression analysis was performed (Table 3). The OR of pancreatitis in dogs with DM were 12.4 times that of dogs without DM (P < .001; 95% CI 4.9–31.6), and the OR of pancreatitis in dogs with HAC were 4.5 times that of dogs without HAC (P < .001; 95% CI 2.1–9.9).
Next, further analyses were performed to compare the proportions of dogs with endocrine diseases and HTG between the pancreatitis and control groups. The frequency of DM with HTG differed between the two groups (P < .001; 95% CI 2.1−13.1; Table 4), whereas those of HAC (P = .176; 95% CI 0.7−6.1) and hypothyroidism (P = .32; 95% CI 1.0−1.0) with HTG did not. The median serum triglyceride concentration of dogs with DM was significantly higher in the pancreatitis group than in the control group (P = .039; Supplementary Figure I).
In addition, the Baron-Kenny approach was used to determine whether the associations between endocrine diseases (DM and HAC) and pancreatitis were mediated by HTG (Table 5). The mediation proportion of the effect of DM on pancreatitis, mediated via HTG, was 0.48, indicating that HTG significantly mediated the relationship between DM and pancreatitis. However, the interaction term (HAC × HTG) was statistically significant, indicating that HTG is a moderator, rather than a mediator, in the relationship between HAC and pancreatitis. In addition, the coefficient of the interaction term was negative. Therefore, in this study, HTG moderated the effect of HAC on the development of pancreatitis in dogs, and more specifically, the effect of HAC on pancreatitis could be less if dogs have concurrent HTG.
Discussion
The present study shows that the prevalence of DM and HAC was significantly higher in dogs with acutely presenting pancreatitis than in dogs without acutely presenting pancreatitis, and dogs with HTG had a higher risk of acutely presenting pancreatitis. In addition, the percentage of dogs with DM and HTG was higher in the pancreatitis group than in the control group, and the median serum triglyceride concentration of dogs with DM was significantly higher in the pancreatitis group than in the control group. Of note, the calculation of the mediation proportions showed that HTG significantly mediated the relationship between DM and acutely presenting pancreatitis. These findings suggest that HTG arising secondarily to DM might be a risk factor for acutely presenting pancreatitis. DM is considered to be a common cause of secondary HTG in dogs, and HTG has also long been thought to be a risk factor for pancreatitis.5,8,25 HTG is a well-known risk factor for pancreatitis in humans, and pancreatitis caused by HTG is termed “hyperlipidemic pancreatitis” in human medicine.26 It has been reported that human patients with hyperlipidemic pancreatitis usually have pre-existing lipid abnormalities caused by poorly controlled DM.27 In humans with type 1 DM, lipoprotein lipase activity is markedly reduced, which leads to increased production and decreased plasma clearance of triglyceride.28 This might also occur in diabetic dogs with HTG.
There are several proposed mechanisms to explain how HTG causes pancreatitis.29 The hydrolysis of triglycerides in and around the pancreas by pancreatic lipase is involved in the induction of pancreatitis by HTG in dogs.28 This leads to a high concentration of plasma-free fatty acids, which are toxic, potentially leading to acinar cell or capillary injury.30 The increased free fatty acids concentration, along with the increased concentration of chylomicrons, can also cause activation of trypsinogen and thereby initiate pancreatitis.30 Furthermore, the accumulation of chylomicrons in the microcirculation disturbs blood flow in pancreatic capillaries, leading to ischemic injury. The pancreas is intrinsically susceptible to ischemia. In human medicine, mutation, variation, and haplotypes of the cystic fibrosis transmembrane conductance regulator and a polymorphism in the tumor necrosis factor promoter are risk factors for hyperlipidemic pancreatitis.31 Moreover, the e4 allele of the apolipoprotein E gene is more frequently identified in human patients with hyperlipidemic pancreatitis.31 These observations imply that hyperlipidemic pancreatitis is a complex disease influenced by the interactions between metabolic, genetic, and environmental factors, including diet and concurrent endocrine diseases. Therefore, it is possible that DM induces secondary HTG, which could lead to acutely presenting pancreatitis. However, a cause-and-effect relationship between HTG arising secondarily to DM and pancreatitis in dogs cannot be clarified from our results because of the nature of retrospective studies. Therefore, the clinical importance of our findings should be interpreted with caution. Further studies are required to confirm the results.
Because this study was retrospective and only HTG was included as a mediator in the Baron-Kenny approach, we did not completely exclude the presence of other mediators of the relationship between DM and pancreatitis. The mediation proportion of HTG in the relationship between DM and pancreatitis was 0.48 (partial mediation), which implies that there is another link between these two conditions. One such possible link is the degree of hyperglycemia. Severe hyperglycemia might directly initiate pancreatitis in dogs.4 In humans, an increased production of reactive oxygen species and lipid peroxidation associated with chronic hyperglycemia has been reported to be involved in the pathogenesis of pancreatitis.32 A recent experimental feline study suggested that hyperglycemia plays a role in the pathogenesis of pancreatitis.33 However, this has not been observed so far in dogs. Although we did not analyze the degree of hyperglycemia or the concentration of fructosamine, future studies will be necessary to investigate this potential direct effect of DM on the development of pancreatitis in dogs.
In contrast to the situation with DM, HTG did not mediate, but instead moderated, the relationship between HAC and pancreatitis. As shown in Table 4, there was no significant difference in the proportion of dogs with HAC and concurrent HTG between the pancreatitis and control groups, and the moderating effect of HTG on the association between HAC and pancreatitis was significant. In addition, the coefficient of the interaction term was negative. Therefore, HTG reduced the effect of HAC on pancreatitis.
The reason for this unexpected result is not known, but there are some possible explanations. Because HAC has been suggested to be a risk factor for pancreatitis, dogs with pancreatitis may be screened for HAC earlier in the clinical course of the disease before they have developed HTG.2,8 In contrast, dogs presenting for a health check may only be screened if they already have overt clinicopathologic abnormalities associated with the disease, such as HTG. In other words, dogs with HAC and pancreatitis are getting diagnosed sooner than those with only HAC. Therefore, the different results obtained regarding DM and HAC could have been due to the differing nature of the diseases, although clinicopathologic abnormalities associated with these diseases are similar.34 Further studies are required to confirm these findings and to compare the differences between these endocrine diseases.
Other variables could also have influenced our findings. There is a difference in the principal lipoprotein class that is affected by DM and HAC.35 DM is principally associated with increased very low-density lipoprotein (VLDL) concentrations, whereas HAC is associated with increased concentrations of both VLDL and low-density lipoprotein in dogs.35 However, endogenous triglyceride is the major lipid component of VLDL, whereas the major component of low-density lipoprotein is cholesterol in dogs.35 As described above, HTG could cause pancreatitis. Therefore, the development of pancreatitis secondary to endocrine diseases could be mediated by VLDL. Although the difference in VLDL concentrations between dogs with DM and HAC is not investigated, it is also possible that this difference in lipoprotein composition might be one of the reasons for the unexpected result. In dogs with uncontrolled HAC, uncontrolled polyphagia may increase the risk of dietary indiscretion, which is an established risk factor for pancreatitis.18,36
In the present study, pancreatitis was diagnosed based on previous studies describing clinical signs, cPL, and ultrasonographic findings.12,13 Dogs with HAC often have a hyperechoic pancreas and increased cPL concentrations.37,38 Although such dogs could have subclinical pancreatitis, cPL and ultrasonographic findings may result in the false-positive diagnosis of pancreatitis in dogs with HAC. We used these diagnostic tests without performing histopathology of the pancreas; therefore, the false-positive diagnosis of pancreatitis in HAC dogs may have influenced our analysis of the role of HTG as a mediator of the relationship between HAC and pancreatitis. Conversely, a previous study reported that ∼8% of 200 sequential dogs without medical records of pancreatitis undergoing postmortem examination had microscopic lesions suggestive of pancreatitis.39 Therefore, a further study based on histopathological analysis of the pancreas is necessary.
No association between the presence of hypothyroidism and the occurrence of pancreatitis was observed in the present study. This result was inconsistent with that of a previous retrospective study.8 We postulate that the low incidence of pancreatitis in dogs with hypothyroidism might have contributed to this negative result. Indeed, only one of the dogs with pancreatitis who presented during the study period was diagnosed with hypothyroidism. In addition to this, we excluded dogs with missing data, which included dogs with a history and clinicopathological findings suggestive of hypothyroidism, along with a low total T4 concentration, but who had not undergone further endocrine tests, particularly a TSH stimulation test. This recruitment criteria might have contributed to the low number of hypothyroid dogs in the study population.
The respective mediation proportions of HTG in the effect of DM on pancreatitis were calculated using the Baron-Kenny approach. The Baron-Kenny approach is an analytic method used to evaluate causal pathways, and it has been used to investigate the causal pathways of various diseases in human medicine.40,41 This result supports a conclusion that the identified association between DM and pancreatitis is mediated by HTG. Nevertheless, a cause-and-effect relationship between endocrine diseases, HTG, and pancreatitis in dogs could not be clearly inferred because of the nature of retrospective studies. Furthermore, the Baron-Kenny approach is used to identify evidence that a variable mediates an effect indirectly, but rarely involves an estimate of the indirect effect itself, instead relying on methods for making statistical inferences that have lower power than some alternatives, including the structural equation approach.42 Therefore, further longitudinal studies are required to verify our results.
Endocrine diseases are slow to progress and are often present for several months before diagnosis. Therefore, the onset of the endocrine diseases might precede the onset of pancreatitis by a significant period of time.8 This assumption might add support to our conclusion that endocrine diseases may be a cause of pancreatitis. In DM, subclinical pancreatitis might also cause progressive pancreatic islet cell destruction, or autoantibodies directed against insulin-secreting cells might promote generalized pancreatic inflammation.4 However, because we also excluded dogs diagnosed with DM after their recovery from pancreatitis, the chance that DM was a consequence of pancreatitis is likely to be low.
This study has several limitations. There are the inherent limitations associated with retrospectively designed studies. In particular, the number of dogs with hypothyroidism was low in both groups, which may be due to underdiagnosis of the condition in this retrospective study. Furthermore, to focus on the role of HTG secondary to DM in the occurrence of pancreatitis, only dogs who did not receive any medications for endocrine diseases were included in the present study. In addition, it is difficult to infer a cause-and-effect relationship between DM and HTG in dogs with pancreatitis because of the retrospective design. Nevertheless, our results support the conclusion that the association between DM and pancreatitis is largely attributable to secondary HTG. To confirm the association, it will be necessary to collect clinical samples before, during, and after resolution of pancreatitis. Although this might be difficult in practice as it is nearly impossible to determine when pancreatitis initially occurs, it might be feasible in long-term observational studies.
We used the same exclusion/selection criteria for the control group. Selection bias can be a serious problem for the estimation of OR in case-control studies. Case-control studies are highly vulnerable to selection bias, particularly in the control group. We adopted several strategies to reduce this bias. First, we used case-control matching, although it could not completely eliminate the possibility of selection bias.43 Previous studies reported that age, breed, and sex can be risk factors for the occurrence of acute pancreatitis and endocrine diseases.44−46 Therefore, we used age- and breed-matched controls, and compared the sex ratio between cases and controls. Second, we selected hospital controls, which may also have been a source of bias. This approach helped us to select controls from the same population, which is crucial for case-control studies. Most of these limitations would be reduced or eliminated by performing a cohort study. However, cohort studies take longer and are harder to conduct because of case selection and controls from exposure status and then followed longitudinally.46
It was not possible to know exactly how long the dogs had been fasted before they underwent diagnostic testing, although we always checked that they had been fasted for at least 12 hr. HTG can be a physiological status in blood samples from nonfasted animals.35 Within 30 min to 2 hr after a meal, chylomicrons become apparent as a transient increase in plasma or serum turbidity, which clears after 6–10 hr.35 To avoid the influence of postprandial lipidemia, previous studies recommend fasting dogs for over 12 hr before assessing plasma lipid concentrations.47 In addition to this, we did not exclude the effect of diet on the occurrence of pancreatitis, although dogs with prediagnosed DM or HAC might have been fed a relatively low-fat commercial or homemade diet according to their owner’s preference. Therefore, we could not completely exclude the possibility that diet affected our results. The body condition score of the dogs was not assessed in this study. Obesity is usually associated with increased serum triglyceride concentration.48 However, because we used breed-matched controls, and body weight was not different between the pancreatitis and control groups, the influence of obesity on the data is not likely to be large. Nevertheless, future studies should be designed to avoid this limitation by additionally matching the body condition score between cases and controls.
In the statistical analysis of our study, HTG was treated as a binary rather than continuous variable. In human medicine, the degree of HTG is important, and patients with uncontrolled severe HTG have the greatest risk for acute and recurrent pancreatitis.26 In the present study, the median serum triglyceride concentration of DM dogs was 5-fold higher in the pancreatitis group (444.5 mg/dL) than in the control group (92.0 mg/dL), and 81% of the DM dogs in the pancreatitis group also had moderate-to-severe HTG (Supplementary Figure I). This could indicate that only severe HTG increases risk for pancreatitis. A further study with a regression model that treats serum triglyceride concentrations as a linear variable is necessary to clearly explain the effect of triglycerides on pancreatitis development in dogs with endocrine diseases, especially DM.
Miniature schnauzers have a high prevalence of primary HTG and might also be predisposed to endocrine diseases, especially DM.5,49 In addition, miniature schnauzers with a history of pancreatitis are five times more likely to have moderate-to-severe HTG (the serum triglyceride concentration >500 mg/dL).50 Therefore, primary HTG in miniature schnauzers may mask the effect of secondary HTG on the risk of pancreatitis. However, we obtained the same results regardless of whether miniature schnauzers were included (data not shown). Only a small proportion of the total population were miniature schnauzers (Supplementary Table I); therefore, we did not perform separate analyses to examine the role of HTG as a mediator between endocrine diseases and pancreatitis in this breed.
Conclusion
In conclusion, our results indicate that the prevalence of DM or HAC in dogs with acutely presenting pancreatitis might be higher. Furthermore, we suggest that HTG resulting from DM might be associated with acutely presenting pancreatitis. Additional large cohort studies are required to confirm these findings.
Contributor Notes
