Predictors of Clinical Remission in Cats with Diabetes Mell i tus

E. Zini, M. Hafner, M. Osto, M. Franchini, M. Ackermann, T.A. Lutz, and C.E. Reusch

Background: Clinical remission is frequent in cats with well-controlled diabetes mellitus, but few studies explored predictors

of this phenomenon.

Hypothesis: Data retrieved from medical records at admission might be valuable to identify likelihood of remission and its

duration in diabetic cats.

Animals: Ninety cats with newly diagnosed diabetes, followed-up until death or remission.

Methods: Retrospective cohort study. Data were collected from records at admission, including history, signalment, phys-

ical examination, haematology, and biochemical profile, and the occurrence and duration of remission, defined as

normoglycemia without insulin for �4 weeks. Predictors of remission were studied with univariate and multivariate logistic

regression. Factors associated with remission duration were analyzed with Kaplan-Meier and Cox proportional hazard models.

Results: Forty-five (50%) cats achieved remission, after a median time of 48 days (range: 8–216). By study end, median

remission duration was 114 days (range: 30–3,370) in cats that died and 151 days (range: 28–1,180) in alive cats. Remission was

more likely with higher age (OR: 1.23, 95% CI: 1.04–1.46; P 5 .01) and less likely with increased serum cholesterol (OR: 0.36,

95%CI: 0.11–0.87; P 5 .04). Remission was longer with higher body weight (HR: 0.65, 95%CI: 0.42–0.99; P 5 .04) and shorter

with higher blood glucose (HR: 1.01, 95% CI: 1.00–1.02; P 5 .02).

Conclusions and Clinical Importance: Age, body weight, cholesterol, and glucose levels are suggested for prediction of re-

mission or its duration in diabetic cats. Older cats developing diabetes may have a better outcome, possibly suggesting a slower

disease progression.

Key words: Feline; Normoglycemia; Prediction; Transient diabetes.

Diabetes mellitus is 1 of the most commonly encoun-tered endocrine diseases in cats.1 Insulin therapy is

the most effective means to achieve glycemic control andto avoid life-threatening complications in diabetic cats.2

Interestingly, 41–84% of affected cats have been reportedto maintain normoglycemia without exogenous insulinwithin weeks to months of beginning therapy.3–7 The termdiabetic remission or transient diabetes is used in catswhen insulin administration can be withdrawn for at least4 consecutive weeks and when, beside restoration of nor-mal glucose concentrations, clinical signs of diabetes havedisappeared.5 Viability of b-cells may not have fully re-covered, as shown by a reduced number of pancreatic isletcells in diabetic cats examined during remission.3 The rea-son why remission occurs in some cats is uncertain.Because feline b-cells are very susceptible to the detrimen-tal effects of excess glucose,8 it is hypothesized that anadequate control of glycemia with insulin may reverseglucose toxicity in the endocrine pancreas.3–7

A small number of investigations addressed diabeticremission in cats to identify predictors of this phenome-

non at the time of admission. At our institution it wasshown that serum concentrations of glucose, fructosa-mine, insulin, glucagon, and insulin growth factor-1 arenot different between diabetic cats that achieved remis-sion and those that did not.4 Another group, by inducingb-cell secretion with a glucagon stimulation test, foundsimilar insulin area under the curve between diabetic catsthat achieved or not remission.3 During remission, gluc-agon-induced insulin secretion was comparable tohealthy cats.3 Recently, a study based on a questionnaireto diabetic cat owners participating in an internet forumshowed that strict glycemic control, administration ofcorticosteroids previous to diagnosis, and absence ofpolyneuropathy were more likely in cats with remission6;age, sex, body weight, renal failure, and hyperthyroidismwere not useful predictors.6 Administration of insulinglargine increased the chance of remission in diabeticcats.7 In a study, we observed that normoglycemia canalso be attained and insulin discontinued in diabetic catswith ketoacidosis.5 Cats with diabetes achieving remis-sion have higher area under the curve of the glucagon-to-insulin ratio than cats with life-long insulin requirement,suggesting a possible role for glucagon in the pathogen-esis of diabetic remission.a

Improving anticipation of remission would be veryhelpful to increase motivation of owners to treat theirdiabetic cats. Achieving remission is advantageousbecause cats benefit from better quality of life and costsof treatment are decreased. Therefore, the aim of thepresent study was to explore factors that at the time ofadmission might predict remission in diabetic cats;we used data retrieved from history, signalment, physi-cal examination, and hematology and biochemicalprofile. In addition, the same factors were studiedto identify possible associations with duration ofremission.

From the Clinic for Small Animal Internal Medicine (Zini, Ha-fner, Reusch), Institute of Veterinary Physiology (Osto, Lutz), andthe Institute of Virology (Franchini, Ackermann), Vetsuisse Faculty,University of Zurich, 8057 Zurich, Switzerland. The study was per-formed at the Clinic for Small Animal Internal Medicine, VetsuisseFaculty, University of Zurich, Switzerland.

Corresponding author: Dr Eric Zini, Clinic for Small AnimalInternal Medicine, Vetsuisse Faculty, University of Zurich, Winter-thurerstrasse 260, 8057 Zurich, Switzerland; e-mail:ezini@vetclinics.uzh.ch.

Submitted April 15, 2010; Revised July 1, 2010; AcceptedAugust 4, 2010.

Copyright r 2010 by the American College of Veterinary InternalMedicine

10.1111/j.1939-1676.2010.0598.x

J Vet Intern Med 2010;24:1314–1321

Materials and Methods

Cats

Medical records of all diabetic cats admitted to the Clinic for

Small Animal Internal Medicine, Vetsuisse Faculty, University of

Zurich (Switzerland), between January 2000 and July 2009 were re-

viewed. Cats were included in the study if they had newly diagnosed

diabetes and were followed up at least until remission was achieved

or, otherwise, until death. Cats with ketoacidosis or concurrent dis-

eases were also included.

From all records, information was retrieved on admission per-

taining history (including administration of corticosteroids and

progestagens in the previous 6 months), signalment, physical exam-

ination, and blood work. Ketoacidosis or any concurrent disease

diagnosed on admission was recorded. The type of insulin adminis-

tered after discharge was also retrieved.

Whether the diabetic cat had achieved clinical remission, that is,

insulin was not required to maintain normoglycemia, and how long

remission had lasted, was obtained from follow-up records. Similar

to previous studies performed by our group, only cats that did not

need insulin for at least 4 consecutive weeks were considered in re-

mission.5,a As a standard procedure at our institution, all diabetic

cats had been reassessed at 1, 2–3, 4–6, 8–12 weeks, and 6 months

following discharge, and thereafter every 3–6 months, regardless of

remission. Additional evaluations had been planned according to

individual needs.

Statistical Analysis

To identify predictors of remission, univariate binary logistic re-

gression analysis was performed with all diabetic cats and the

following factors used as dichotomous variables: previous adminis-

tration of corticosteroids and progestagens, sex (male or female),

breed (pure- or cross-bred), increase of leukocyte count, serum cho-

lesterol, total protein, creatinine, urea, potassium, bilirubin and

lipase, decrease of hematocrit, serum albumin and potassium, and

presence of ketoacidosis or of concurrent disease. Increase and de-

crease of above factors was relative to reference ranges established

at our institution (Clinical Laboratory, Vetsuisse Faculty, Univer-

sity of Zurich, Switzerland). The type of insulin administered, either

porcine insulin zinc suspensionb or insulin glargine,c was also in-

cluded in the analysis; cats that received 1 of the 2 insulin types but

died before discharge were not considered. Glucose and fructosa-

mine concentrations that are typically increased in untreated

diabetic cats were investigated as continuous variables. Age and

body weight were also studied as continuous variables. Factors that

at univariate analysis yielded Po.25 were further used to assess

their confounding effect with multivariate binary logistic regression.

Factors whose results were not available in more than 20% of cats

were excluded from the multivariate analysis to preserve the fit of

the model.

In the group of cats that had insulin discontinued, analysis was

conducted to identify factors associated with duration of remission.

The influence of above dichotomous variables was studied with

Kaplan-Meier product-limit followed by log rank test; only dichot-

omous variables represented by at least 10 cases in either category

were considered. The influence of continuous variables was studied

with univariate Cox proportional hazard model. Factors with

Po.25 were further used to assess their confounding effect using

the Cox proportional hazard model. Cats in remission were cen-

sored if they were still alive by the study end.

Significance was considered for Po.05. Statistical analysis was

conducted with software.d,e

Results

Cats

Of 275 diabetic cats admitted during the study period,90 met the inclusion criteria and were included in theanalysis. Forty-five (50%) of the 90 diabetic catsachieved remission and 45 (50%) did not.

Cats with Diabetic Remission

In the group that achieved remission, median age atdiagnosis was 12 years (range: 2–17). Thirty-two (71%)cats were neutered male and 13 (29%) spayed female.Thirty (67%) cats were cross-bred domestic short- orlonghair, and 11 (33%) were pure-bred, including 3 Per-sian, 2 each Maine Coon and Siamese, and 1 eachAbyssinian, Birman, Burmese, and Ragdoll; breed wasnot recorded in 4 cats. Median body weight was 5.2 kg(range: 2.5–8.4). Corticosteroids or progestagens hadbeen administered to 5 cats, up to 3 months beforeadmission.

On admission median serum glucose was 419mg/dL(range: 248–1,251; reference range: 90–180), andfructosamine was 588 mmol/L (range: 371–984; referencerange: 200–300). Ketoacidosis was diagnosed in 13(29%) cats and concurrent diseases were documented in20 (44%) cats, including 3 each with lung disease or sus-pected pancreatitis, 2 each with cystitis, felineimmunodeficiency virus or skin disease, and 1 each withbone fracture, cholangiohepatitis, enteritis, hyper-thyroidism, lymphoma, pure red cell aplasia, renalfailure, or stomatitis. The frequency of abnormal bloodresults is summarized in Table 1.

After discharge, 21 (47%) cats were treated with insu-lin glargine and 23 (53%) with porcine insulin zincsuspension; in 1 cat the type of insulin is unknown. Ad-ditional therapy was administered to treat concurrentdiseases (data not shown). Median time from dischargeto remission was 48 days (range: 8–216); 25% of catsachieved remission within 27 days, 75% within 102 days,and only 1 cat after 6 months (Fig 1). By the end of thestudy, median remission duration was 114 days (range:30–3,370) in cats that were followed up until death (n 5

15) and 151 days (range: 28–1,180) in cats that were stillalive (n 5 30). Six of the 45 (13%) cats with remission didnot require insulin for more than 1,000 days; 1 of themfor more than 9 years. Overall, insulin therapy needed tobe resumed to control hyperglycemia in 13 (29%) cats,including 6 of the cats that died and 7 of those that werestill alive at the study end. None of these 13 cats experi-enced a further period of clinical remission.

Cats without Diabetic Remission

In the group that did not achieve diabetic remission,median age at diagnosis was 11 years (range: 2–18); in 1cat age was unknown. The cats consisted of 30 (67%)neutered male and 15 (33%) spayed female. Twenty-nine(71%) cats were domestic short- or longhair, and 12(29%) were pure-bred, including 2 each Maine Coon andSiamese, and 1 each Persian, Havana Brown, Birman,

1315Remission from Diabetes in Cats

Burmese, Chartreux, Norwegian Forest, Oriental short-hair, and Russian Blue; breed was not recorded in 4 cats.Median body weight was 4.8 kg (range: 3.0–9.8); body

weight was not available in 1 cat. Corticosteroids or pro-gestagens were administered to 5 cats and up to 4 monthsbefore admission.

Table 1. Characteristics of diabetic cats with and without clinical remission, and univariate analysis of predictors ofdiabetic remission.

Variablea Reference Range Cats with Remission Cats without Remission OR 95% CI P-Value

Age 1.13 0.99–1.28 .07

Sex

Neutered male 32 (71%) 30 (67%) 1.23 0.50–3.01 .65

Spayed female 13 (29%) 15 (33%)

Breed

Pure bred 11 (33%) 12 (29%) 0.89 0.34–2.33 .81

Cross bred 30 (67%) 29 (71%)

Body weight 0.99 0.75–1.30 .92

Previous administration of corticosteroids or progestagens

Yes 5 (11%) 1.00 0.27–3.73 .00

No 40 (89%) 40 (89%)

Hematocrit 33–45%

Decreased 12 (27%) 16 (37%) 0.59 0.25–1.50 .23

Not decreased 33 (73%) 27 (63%)

Leukocyte count 4.6–12.8 � 103/mLIncreased 22 (50%) 23 (55%) 0.83 0.35–1.93 .66

Not increased 22 (50%) 19 (45%)

Glucose concentration 90–180mg/dL 1.00 0.99–1.00 .95

Fructosamine concentration 200–300mmol/L 1.00 0.99–1.00 .70

Cholesterol concentration 100–263mg/dL

Increased 15 (34%) 24 (55%) 0.43 0.18–1.02 .05

Not increased 29 (66%) 20 (45%)

Total protein concentration 64–80 g/L

Increased 10 (23%) 10 (22%) 1.03 0.38–2.79 .95

Not increased 34 (77%) 35 (78%)

Albumin concentration 30–40 g/L

Decreased 6 (14%) 8 (18%) 0.73 0.23–2.31 .59

Not decreased 38 (86%) 37 (82%)

Creatinine concentration 1.1–1.7mg/dL

Increased 7 (16%) 11 (25%) 0.57 0.20–1.63 .29

Not increased 37 (84%) 33 (75%)

Urea concentration 21–35mg/dL

Increased 12 (28%) 21 (47%) 0.43 0.18–1.04 .06

Not increased 32 (72%) 24 (53%)

Potassium concentration 3.8–5.4mEq/L

Increased 4 (10%) 8 (18%) 0.49 0.14–1.76 .27

Not increased 38 (90%) 37 (82%)

Potassium concentration 3.8–5.4mEq/L

Decreased 7 (17%) 11 (24%) 0.62 0.21–1.78 .37

Not decreased 35 (83%) 34 (76%)

Bilirubin concentration o.2mg/dL

Increased 14 (38%) 20 (57%) 0.46 0.18–1.17 .10

Not increased 23 (62%) 15 (43%)

Lipase activity 8–26U/L

Increased 26 (59%) 28 (65%) 0.77 0.33–1.85 .56

Not increased 18 (41%) 15 (35%)

Ketoacidosis

Yes 13 (29%) 17 (38%) 0.67 0.28–1.62 .37

No 32 (71%) 28 (62%)

Concurrent disease

Yes 20 (44%) 24 (53%) 0.70 0.31–1.61 .40

No 25 (56%) 21 (47%)

Type of insulin

Insulin glargine 21 (47%) 8 (31%) 2.05 0.74–5.71 .16

Porcine insulin zinc suspension 23 (53%) 18 (69%)

CI, confidence interval; OR, odds ratio.aAge, body weight, glucose, and fructosamine do not have frequencies listed because analyzed as continuous variables.

1316 Zini et al

On admission median serum glucose was 436mg/dL(range: 243–990) and fructosamine was 613 mmol/L(range: 435–968). Ketoacidosis was diagnosed in 17(38%) cats and concurrent diseases were documented in24 (53%) cats, including 3 each with cholangiohepatitis,hypertrophic cardiomyopathy or lung disease, 2 eachwith cystitis, renal failure, or stomatitis, and 1 each withacromegaly, enteritis, epilepsy, histiocytosis, idiopathichypercalcemia, lymphoma, megacolon, skin disease, orsuspected pancreatitis (Table 1).After discharge, 8 (31%) cats were treated with insulin

glargine and 18 (69%) with porcine insulin zinc suspen-sion. Twelve cats received regular insulin on admissionbecause of ketoacidosis and did not survive to discharge.Seven additional cats died before discharge; 3 of themreceived insulin glargine, lente insulin, or regular insulin,and 4 died previous to insulin therapy. Additional ther-apy was administered to treat concurrent diseases.

Predictors of Diabetic Remission

Factors retrieved from medical records on admissionwere studied to assess their role as predictors of diabeticremission. By univariate logistic regression analysis,higher age and the use of insulin glargine were morelikely in diabetic cats that achieved remission, whereaslow hematocrit and increased cholesterol, urea andbilirubin were more likely in diabetic cats that did notachieve remission (Table 1). Sex, breed, body weight,administration of corticosteroids, or progestagens, leu-kocyte count, serum glucose, fructosamine, total protein,albumin, creatinine, potassium concentrations, lipase ac-tivity, ketoacidosis, and concurrent diseases did notpredict remission in our population of cats.The multivariate logistic regression model was con-

structed with the continuous variable age and thedichotomous variables hematocrit, cholesterol, urea,and bilirubin. Type of insulin was not included becausemore than 20% of cats did not survive to discharge.Based on multivariate analysis, age and serum choles-terol were independent predictors of diabetic remission

(Po.05) (Table 2). In particular, increased cholesteroldecreased the chance of remission by almost 65%, andper year of age chance of remission increased by approx-imately 25%. Approximately 70% of cats that were olderthan 12 years at diagnosis achieved diabetic remission,less than 50% of cats that were between 10 and 12 years,and only 30–40% of cats below 9 years. The Hosmer-Lo-meshow test was not significant (P 5 .34) suggesting fit ofthe multivariate model.

Factors Associated with Duration ofDiabetic Remission

Factors retrieved from medical records on admissionwere investigated to assess possible associations with theduration of diabetic remission. Results of Kaplan-Meieranalysis for dichotomous variables and of univariate Coxproportional hazard model analysis for continuous vari-ables are reported in Table 3; higher body weight wasassociated with longer remission, whereas decreasedhematocrit, higher serum glucose concentration, and in-creased serum lipase activity were associated with shorterremission. Age, sex, breed, leukocyte count, serumfructosamine, cholesterol, total protein, urea and bili-rubin, ketoacidosis, concurrent diseases, and type ofinsulin were not linked to the duration of remission. Theeffects of administration of corticosteroids or pro-gestagens and serum albumin, creatinine and potassiumwere not assessed because we observed less than 10 casesper category.

Results of the multivariate Cox proportional hazardmodel constructed with the dichotomous variables hem-atocrit and serum lipase, and the continuous variablesbody weight and serum glucose are reported in Table 4.With the adopted model, body weight and serum glucoseremained independently associated with the duration ofdiabetic remission (Po.05). In particular, for a 1 kg in-crease of body weight, duration of remission increased by

70

80

90

100

40

50

60

10

20

30

% o

f cat

s in

rem

issi

on

0 25 50 75 100 125 150 175 200 225 2500

Days after diagnosis

Fig 1. Diabetic cats achieving clinical remission. Cumulative per-

centage of cats that did not require insulin to maintain

normoglycemia over time.

Table 2. Multivariate analysis of predictors ofdiabetic remission.

Variable

Reference

Range

Chance of Diabetic

Remission

OR 95% CI P-Value

Age 1.23 1.04–1.46 .01

Hematocrit 33–45%

Decreased 0.48 0.14–1.69 .26

Not decreased

Cholesterol concentration 100–263mg/dL

Increased 0.36 0.11–0.87 .04

Not increased

Urea concentration 21–35mg/dL 0.59 0.18–2.00 .40

Increased

Not increased

Bilirubin concentration o.2mg/dL

Increased 0.49 0.16–1.47 .20

Not increased

CI, confidence interval; OR, odds ratio.

1317Remission from Diabetes in Cats

approximately 35%, and for a 1mg/dL increase of glu-cose in the hyperglycemic range, duration of remissiondecreased by approximately 1% (Figs 2 and 3). The like-lihood ratio test of the model yielded an overall w2

difference significant at Po.001, indicating that the vari-ables accurately contributed to explaining the durationof remission.

Discussion

In the present study 45 of 90 (50%) diabetic catsachieved clinical remission, and the large majority of

them within 6 months from diagnosis, similar to previousreports.3–7 Despite remission being frequent in cats, fewstudies investigated the factors that could predict this fa-vorable outcome already at the time of 1st presentationto the clinic.3–7 Here, using commonly measured param-eters retrieved on admission from history, signalment,physical examination, hematology, and biochemical pro-file, we identified predictors of remission and of itsduration. In particular, remission was more likely withhigher age, and less likely with elevated serum choles-terol. Remission was longer with higher body weight,and shorter with higher serum glucose.

Table 3. Univariate analysis of factors associated with duration of diabetic remission. The number of cats isprovided in parentheses for each variable.

Variablea,b Reference Range

Duration of Diabetic Remission

Median (days) HR 95% CI P-Value

Age 0.99 0.85–1.16 .92

Sex

Neutered male (32) 1240 1.69 0.54–4.87 .39

Spayed female (13) 1380

Breed

Pure-bred (11) 3370 0.34 0.10–1.12 .38

Cross-bred (30) 1240

Body weight 0.67 0.46–0.98 o.01

Hematocrit 33–45%

Decreased (12) 150 3.03 1.27–19.8 .02

Not decreased (33) 1380

Leukocyte count 4.6–12.8 � 103/mLIncreased (22) 1240 2.17 0.80–6.25 .27

Not increased (22) 1380

Glucose concentration 90–180mg/dL 1.01 1.00–1.02 o.01

Fructosamine concentration 200–300mmol/L 1.00 0.99–1.00 .50

Cholesterol concentration 100–263mg/dL

Increased (15) 1020 1.81 0.68–6.67 .29

Not increased (29) 1380

Total protein concentration 64–80 g/L n.a. 1.35 0.34–5.88 .64

Increased (10)

Not increased (34) 1240

Urea concentration 21–35mg/dL

Increased (12) 1240 1.20 0.39–4.01 .72

Not increased (32) 3370

Bilirubin concentration o.2mg/dL

Increased (14) 1240 1.28 0.41–4.34 .64

Not increased (23) 1020

Lipase activity 8–26U/L

Increased (17) 150 4.17 1.61–16.7 .02

Not increased (27) 1020

Ketoacidosis

Yes (13) 1240 0.50 0.17–1.39 .27

No (32) 1020

Concurrent disease

Yes (20) 1380 1.52 0.57–4.35 .73

No (25) 1240

Type of insulin

Insulin glargine (21) n.a. 0.88 0.27–2.70 .79

Porcine insulin zinc suspension (23) 1240

CI, confidence interval; HR, hazard ratio; n.a., not available.aKaplan-Meier analysis is used for dichotomous variables; variables with less than 10 cases in 1 category are not assessed (ie, previous

administration of corticosteroids and progestagens, serum albumin, creatinine, and potassium).bUnivariate Cox proportional hazardmodel analysis is used for continuous variables, including age, body weight, glucose, and fructosamine.

1318 Zini et al

Higher age at admission was associated with an aug-mented chance of diabetic remission. Specifically, foreach additional year of age at 1st diagnosis, remissionwas 25% more likely. This finding may be unexpectedbased on the notion that aging is associated with de-creased b-cell function in different mammalianspecies.9,10 However, a study demonstrated that type 2diabetic humans first diagnosed after 65 years of agepresent with a less severe disease course and less severemetabolic deterioration. At baseline and during follow-up, type 2 diabetics that were older than 65 years at di-agnosis had lower fasting blood glucose and glycatedhemoglobin concentrations.11 The reason why diabetesin older humans evolves more slowly has not been clar-ified. Also in type 1 diabetic patients, younger age atdiagnosis was associated with a faster progression to in-sulin dependency.12,13 It cannot be excluded that thepositive association between remission and age observedin cats is because of owners being disposed to providebetter care in older animals.Higher than normal cholesterol reduced the chance of

remission by almost 65% in diabetic cats. Most studies inhumans linking hyperlipidemia to type 2 diabetes had fo-cused on the role of triglycerides and free fatty acids, butrecent in vivo studies with rodents have shown that in-

creased cholesterol concentrations can also impair b-cellfunction or viability.14,15 Further, isolated pancreaticislets had increased content of cholesterol, and itsdepletion was followed by restoration of insulin secre-tion.14 In another transgenic mice model characterizedby hypercholesterolemia, besides impaired insulin secre-tion, the number and size of pancreatic islets werereduced by more than 50%.15 Therefore, increased se-rum cholesterol can exert a direct toxic effect on b-cells.It is possible that hypercholesterolemia also plays a pri-mary role in the progression of diabetes in cats,eventually preventing the recovery of b-cell function orviability. Because in the present study the concentrationsof serum triglycerides and of free fatty acids werenot available, it cannot be excluded that these lipid me-tabolites had a confounding effect on the chance ofremission in our cat population. However, sustainedhypertriglyceridemia and increased free fatty acid con-centrations did not impair insulin secretion in a recentstudy with healthy cats.8

Based on univariate analysis, diabetic cats treated withinsulin glargine tended to have an increased chance ofremission. Because many cats did not survive to dis-charge, hence not receiving insulin glargine or porcineinsulin zinc suspension, the multivariate model wasnot performed to assess whether the former indepen-dently predicts remission. A higher probability ofdiabetic remission was previously reported in cats treatedwith insulin glargine compared with porcine lenteinsulin.7

The duration of diabetic remission varied widely, withsome cats not requiring treatment for few months andothers for many years. More than 10% of cats remainedin remission for more than 1,000 days and 1 cat forlonger than 9 years. Remission lasting more than 1,000days seems to be uncommon in cats.3,6 Insulin therapywas resumed in 13 of the 45 (29%) cats that had experi-enced remission. Likewise, in 1 report 38% of diabeticcats in remission had insulin started again to maintainnormoglycemia.6

Higher body weight was associated with long-lastingdiabetic remission; for a 1 kg increase of body weight at

Table 4. Multivariate analysis of factors associated withduration of diabetic remission.

Variable

Reference

Range

Duration of Diabetic

Remission

HR 95% CI P-Value

Body weight 0.65 0.42–0.99 .04

Glucose concentration 90–180mg/dL 1.01 1.00–1.02 .02

Hematocrit 33–45%

Decreased 2.94 0.86–10.1 .08

Not decreased

Lipase activity 8–26U/L

Increased 1.61 0.42–6.25 .49

Not increased

CI, confidence interval; HR, hazard ratio.

6.0

6.5

7.0

5.0

5.5

3.5

4.0

4.5

Bod

y w

eigh

t (kg

)

3.0≤ 60 ≤ 120 ≤ 180 ≤ 240 > 240

Duration of remission (days)

Fig 2. Duration of clinical remission and body weight at diagnosis

in diabetic cats followed-up until death. Box and whisker plots are

reported.

1250

1500

750

1000

250

500

Glu

cose

(m

g/d

L)

0≤ 60 ≤ 120 ≤ 180 ≤ 240 > 240

Duration of remission (days)

Fig 3. Duration of clinical remission and glucose concentrations at

diagnosis in diabetic cats followed-up until death. Box and whisker

plots are reported.

1319Remission from Diabetes in Cats

admission remission length increased by almost 35%.Obese cats are predisposed to insulin resistance, which inturn poses an excessive demand to b-cells leading to theirexhaustion.16 In this setting, we assumed that the chanceof remission should be reduced. Unfortunately, it wasnot possible to determine whether our cats with higherbody weight had been obese or just were larger cats asbody score was not reported in most medical records.However, body weight was not significantly decreased atthe time of remission compared with body weight at 1stdiagnosis (data not shown). Thus, it seems unlikely thatremission occurred due to a reduction of body weight,which by itself should improve insulin sensitivity and, inturn, b-cell function or viability. Because cats lose bodyweight with untreated diabetes, it cannot be excludedthat the observed association with body weight reflectschronicity of diabetes. Lighter compared with heaviercats might have experienced subclinical diabetes for alonger period on presentation. The lighter cats couldhave been overweight or obese earlier in life and the in-sulin resistance they experienced might have reducedtheir ability to sustain remission.Diabetic cats in remission presenting with higher ini-

tial blood glucose concentrations had insulin withdrawnfor a shorter time. Because feline pancreatic islets are ex-tremely susceptible to the toxic effects of excess glucose,8

it may be assumed that cats with higher glucose concen-trations at initial diagnosis had more severe b-celldamage. The remaining pool of functional b-cells maythus have been more limited in these cats, hence reducingthe chance of lasting remission. Typically, measurementof serum fructosamine is indicated to better assess aver-age blood glucose during the weeks before admission.17

Different from blood glucose, however, fructosaminewas not associated with longer remission. Thus, furtherstudies are necessary to confirm the role of blood glucoseconcentrations on admission.There are some limitations of our study that need to be

mentioned. The retrospective nature of the investigationoccasionally resulted in incomplete clinical data records.However, prospective longitudinal studies spanning overseveral years are difficult in veterinary medicine. Hence,including a relatively large number of cases is easier toperform in retrospective analysis.The role of feeding a specific diet, which is important

to improve the control of glycemia in diabetic cats,18 wasnot assessed in the present series because cats were fedvarious types or were switched to another in many cases.Thus, the potential confounding effect of a particulardiet or formulation cannot be ruled out.Remission applies to cats that achieve normoglycemia

and disappearance of clinical signs of diabetes, withoutinsulin. According to recent studies performed at our in-stitution,5,a we considered a cat to be in remission afterinitial treatment if insulin was not required for at least 4weeks. Based on the latest American Diabetes Associa-tion statement, remission in humans with either type 1 or2 diabetes is defined as attaining normoglycemia in theabsence of active pharmacologic therapy for more than 1year.19 Because of the shorter lifespan of cats, this cut-offdoes not seem suitable for the feline species.

Footnotes

a Tschuor F, Furrer D, Kaufmann K, et al. Intravenous arginine

stimulation test in cats with transient and nontransient diabetes

mellitus. J Vet Intern Med 2006;20:725–726 (abstract)b Caninsulin, Intervet International BV, Boxmeer, the Netherlandsc Lantus, Sanofi Aventis (Suisse) SA, Meyrin, SwitzerlanddGraphPad Prism version 4.0, GraphPad Software, San Diego, CAe SPSS version 11.0, SPSS Inc, Chicago, IL

Acknowledgments

This study was partially supported by a grant fromNestle Purina PetCare.

References

1. Panciera DL, Thomas CB, Eicker SW, et al. Epizootiologic

patterns of diabetes mellitus in cats: 333 cases (1980–1986). J Am

Vet Med Assoc 1990;197:1504–1508.

2. Michiels L, Reusch CE, Boari A, et al. Treatment of 46 cats

with porcine lente insulin: A prospective, multicentre study. J Feline

Med Surg 2008;10:439–451.

3. Nelson RW, Griffey SM, Feldman EC, et al. Transient clin-

ical diabetes mellitus in cats: 10 cases (1989–1991). J Vet InternMed

1999;13:28–35.

4. Alt N, Kley S, Tschuor F, et al. Evaluation of IGF-1 levels in

cats with transient and permanent diabetes mellitus. Res Vet Sci

2007;83:331–335.

5. Sieber-Ruckstuhl NS, Kley S, Tschuor F, et al. Remission of

diabetes mellitus in cats with diabetic ketoacidosis. J Vet InternMed

2008;22:1326–1332.

6. Roomp K, Rand J. Intensive blood glucose control is safe

and effective in diabetic cats using home monitoring and treatment

with glargine. J Feline Med Surg 2009;11:668–682.

7. Marshall RD, Rand JS, Morton JM. Treatment of newly di-

agnosed diabetic cats with glargine insulin improves glycaemic

control and results in higher probability of remission than prota-

mine zinc and lente insulins. J Feline Med Surg 2009;11:683–691.

8. Zini E, Osto M, Franchini M, et al. Hyperglycaemia but not

hyperlipidaemia causes beta cell dysfunction and beta cell loss in the

domestic cat. Diabetologia 2009;52:336–346.

9. Szoke E, Shrayyef MZ, Messing S, et al. Effect of aging on

glucose homeostasis: Accelerated deterioration of beta-cell function

in individuals with impaired glucose tolerance. Diabetes Care

2008;31:539–543.

10. Perfetti R, Rafizadeh CM, Liotta AS, et al. Age-dependent

reduction in insulin secretion and insulin mRNA in isolated islets

from rats. Am J Physiol 1995;269:E983–E990.

11. Kolb H, Schneider B, Heinemann L, et al. Type 2 diabetes

phenotype and progression is significantly different if diagnosed

before versus after 65 years of age. J Diabetes Sci Technol 2008;2:

82–90.

12. Wilson DM. Progress in the treatment of childhood diabetes

mellitus and obesity. J Pediatr Endocrinol Metab 2002;15:

S745–S749.

13. Achenbach P, Warncke K, Reiter J, et al. Type 1 diabetes

risk assessment: Improvement by follow-up measurements in young

islet autoantibody-positive relatives. Diabetologia 2006;49:

2969–2976.

14. Hao M, Head WS, Gunawardana SC, et al. Direct effect of

cholesterol on insulin secretion: A novel mechanism for pancreatic

beta-cell dysfunction. Diabetes 2007;56:2328–2338.

1320 Zini et al

15. Ishikawa M, Iwasaki Y, Yatoh S, et al. Cholesterol accumu-

lation and diabetes in pancreatic beta-cell-specific SREBP-2

transgenic mice: A new model for lipotoxicity. J Lipid Res

2008;49:2524–2534.

16. Appleton DJ, Rand JS, Sunvold GD. Insulin sensitivity de-

creases with obesity, and lean cats with low insulin sensitivity are at

greatest risk of glucose intolerance with weight gain. J Feline Med

Surg 2001;3:211–228.

17. Reusch CE, Haberer B. Evaluation of fructosamine in

dogs and cats with hypo- or hyperproteinaemia, azotaemia,

hyperlipidaemia and hyperbilirubinaemia. Vet Rec 2001;148:

370–376.

18. Bennett N, Greco DS, Peterson ME, et al. Comparison of a

low carbohydrate-low fiber diet and a moderate carbohydrate-high

fiber diet in the management of feline diabetes mellitus. J Feline

Med Surg 2006;8:73–84.

19. Buse JB, Caprio S, Cefalu WT, et al. Consesus statement:

How do we define cure of diabetes? Diabetes Care 2009;32:2133–

2135.

1321Remission from Diabetes in Cats