The work was carried out in Veterinary Medical Teaching Hospital, National Chung-Hsing University, Taichung, Taiwan.
Prognostic Factors and a Prognostic Index for Cats with Acute Kidney Injury
Version of Record online: 10 APR 2012
Copyright © 2012 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 26, Issue 3, pages 500–505, May-June 2012
How to Cite
Lee, Y.-J., Chan, J.P.-W., Hsu, W.-L., Lin, K.-W. and Chang, C.-C. (2012), Prognostic Factors and a Prognostic Index for Cats with Acute Kidney Injury. Journal of Veterinary Internal Medicine, 26: 500–505. doi: 10.1111/j.1939-1676.2012.00920.x
- Issue online: 2 MAY 2012
- Version of Record online: 10 APR 2012
- Manuscript Accepted: 29 FEB 2012
- Manuscript Revised: 4 JAN 2012
- Manuscript Received: 16 SEP 2011
- Case-fatality rate;
- Renal failure
The clinical manifestations of acute kidney injury (AKI) range from mild to fatal in cats; however, prognosis factors have been rarely studied.
To find the clinical factors significantly correlated with the outcome among cats with AKI and to develop a simple prognostic index.
Seventy cats with AKI were recruited.
Demographic and clinicopathological data obtained from 70 cats with AKI were retrospectively collected. Student's t-test or Mann–Whitney U-test and Pearson chi-square test or Fisher's exact were applied to determine the factors associated with survival in cats with AKI. Using logistic regression, the statistically significant factors associated with prognosis were identified and a new prediction model was generated.
The overall case fatality rate was 64% (45/70). The results showed that nonsurviving cats had significantly lower levels of PCV, WBC, RBC, LDH and albumin, a lower albumin/globulin ratio, lower blood glucose, and a reduced body temperature, as well as being older. Serum urea and creatinine concentrations were not statistically significant as prognostic factors, but a decrease in these 2 variables in 3 days was significantly related to a reduction in death. A summary prognostic index including body temperature and LDH and albumin concentrations had area under the receiver-operating characteristic curve (AUROC) for predicting death of 0.86 (P < .05) and a cut-off value of 0.82, a sensitivity of 77% and a specificity of 90%.
The prognosis in cats with AKI is quite different from that found for human and dogs.
area under the receiver operating characteristic curve
decrease in concentration of creatinine in 3 days
decrease in concentration of SUN in 3 days
receiver operating characteristic curve
serum urea nitrogen
white blood cell
Acute kidney injury (AKI), which is characterized by a sudden decrease in renal function and a loss of capacity to regulate fluids and electrolytes, excrete wastes and concentrating urine, is an important disease in both human and veterinary medicine. Case fatality rate due to AKI is relatively high among humans (50% despite dialysis)[2, 3] as well as among dogs (60%).[4, 5] The clinical manifestations of AKI range from mild to fatal. In such circumstances, prognostic factors that are able to predict death and prevent disease progression are important to manage AKI.
The survival rate for AKI is not only determined by renal dysfunction but also is related to the presence of other complications. In humans with AKI, factors associated with death include age,[7, 8] gender, oliguria,[8, 9] sepsis,[9, 10] a low concentration of albumin, fluid overload, and an increase in serum creatinine.[13-15] However, in dogs with AKI, death is associated with an increase in serum creatinine,[4, 16] hypocalcemia, proteinuria, a higher phosphorus concentration, and greater age. Therefore, these factors could be considered concurrently to evaluate the prognosis of AKI in cats.
Despite its importance in feline medicine, reports exploring AKI in cats are rare. Previous studies have mostly been case reports or studies based on the cases with a single cause.[18, 19] In these circumstances, the prognostic factors related to AKI in cats have been largely ignored. To our knowledge, only a single retrospective study of 32 cats with intrinsic AKI has attempted to explore the risk factors in cats associated with AKI.
In the previous study, the poor prognostic factors for cats with AKI consisted of elevated serum potassium and bicarbonate concentrations together with a reduction in albumin concentration. These prognosis variables in cats are quite different from those described for dogs and humans. Additionally, the etiology and management of AKI seem to differ from one geographic region to another, which might affect prognostic factors. Therefore, in-depth studies are needed to obtain further information that might be helpful when carrying out clinical evaluations on cats with AKI.
The objective of this study was to evaluate the demographic and clinicopathological factors that might be related to the outcome of AKI in cats when the cat presents at an animal hospital. After these significant factors have been identified, they can be used to develop a summary prognostic index to help with the clinical evaluation of cats with AKI.
Material and Methods
The studied individuals with AKI were selected from cats that had attended the Veterinary Medical Teaching Hospital of the National Chung-Hsing University from 1998 to 2006. The retrospective dataset was collected using medical records. To identify cats with AKI, cases with an elevated serum creatinine concentration (>2 mg/dL) and serum urea nitrogen (SUN) (>35 mg/dL) were recruited as suspected AKI cases. However, cats post renal or chronic kidney diseases were then excluded. The cats with any urinary obstructive disease, such as calculi in the urinary tract, rupture of urinary bladder, cystitis or feline lower urinary tract disease, were considered postrenal failure cases. Furthermore, cases with one of following clinical signs, persistent azotemia (a clinical history more than 14 days), chronic polyuria, polydipsia or small kidneys, were determined to be chronic kidney disease cases. Besides, through ultrasonic images, cats with suspected renal mass, but finally classified as neoplastic disease or feline infectious peritonitis (FIP), were excluded in this study.
Demographic information was obtained from the medical records and included age, sex, body weight, and BT. Clinical pathological data were collected, including Hb, PCV, WBC, serum creatinine, SUN, SUN/creatinine ratio, AST, ALT, LDH, CK, albumin, A/G ratio, calcium, phosphorus, glucose, serum potassium, sodium, and chloride. Clinical symptoms such as vomiting, diarrhea, anorexia, oliguria/auria, and the duration of hospitalization were also recorded. Additionally, the decrease in SUN (DSUN) and creatinine (DCr) in 3 days was calculated.
Variable data with a normal distribution that were assessed by Shapiro-Wilk test were expressed as means with standard deviation (SD) to show data variation. The Student's t-test was used to compare mean values between 2 groups. Non-normally distributed continuous data were reported as medians, the limits of the overall range and were statistically analyzed by the Mann–Whitney U-test. Categorical data were reported as proportions and the proportional difference between groups were evaluated by Pearson chi-square test or Fisher's exact test as appropriate. Univariate logistic regression analysis was applied first to determine the possible risk factors associated with death due to AKI. After these significant factors were verified, multiple logistic regression analysis using backward selection at the chosen critical level (P < .05) was then carried out to incorporate these factors into the model and generate the best equation for predicting death. The best equation was used to calculate log odds and was used as the summarized prognostic index to predict death due to AKI. A receiver operating characteristic (ROC) curve was constructed using the relationship between the prognostic index and death due to AKI. The area under the receiver operating characteristic curve (AUROC) was calculated to evaluate the diagnostic performance of the prognostic index. For all analyses, values of P ≤ .05 were considered statistically significant. All analyses were performed with commercial software, namely SPSS version 16 for Windows (SPSS Company, Chicago, IL).
Data from 70 cats, identified using the inclusion and exclusion criteria, were analyzed. The 20-day case fatality rate was 64% (45/70). Compared with the AKI cats in the survival group, the nonsurvival (death) group cats were significantly older (median: 2 years versus 6 years old). There was no significant difference between these 2 groups with regard to the body weight. Body temperature was lower in nonsurvival group (median of survivors: 38.2°C; median of nonsurvivors: 37.3°C). No other clinical signs were associated with death due to AKI in cats (Table 1).
|Variable||Survival Group||Nonsurvival Group||P Valuea|
|Age (year)||2 (2)b (n = 15)||6 (4)b (n = 33)||<.001|
(n = 25)
(n = 40)
(n = 24)
(n = 31)
|Sex (male) (%)||64 (n = 26)||63.6 (n = 44)||.98|
|Vomiting (%)||52 (n = 25)||37.1 (n = 35)||.25|
|Diarrhea (%)||8 (n = 25)||5.7 (n = 35)||.73|
In terms of clinical pathological factors, the nonsurvival group had significant lower values for PCV (50% in survivors versus 43.3% in nonsurvivors), WBC (26700 cells/μL in survivors versus 17700 cells/μL in nonsurvivors), RBC (10.7% in survivors versus 8.0% in nonsurvivors), LDH (439 IU/L in survivors versus 252.5 IU/L in nonsurvivors), albumin (4.2 g/dL in survivors versus 3.6 g/dL in nonsurvivors), A/G ratio (1.26 in survivors versus 1.05 in nonsurvivors) and blood glucose (198 g/dL in survivors versus 147 g/dL in nonsurvivors) (Table 2). However, the values for SUN (194 mg/dL in survivors versus 238 mg/dL in nonsurvivors; P = .14), serum creatinine (6.6 mg/dL in survivors versus 9.5 mg/dL in nonsurvivors; P = .07) and electrolytes (calcium, phosphorus, sodium, potassium and chloride), as well as the SUN/creatinine ratio, among the nonsurvival group were not significantly different from those of members of the survival group.
|Variable||Survival Group||Nonsurvival Group||P Valuec|
|Hbb (g/dL)||14.9 (2.9); n = 21||12.1 (3.1); n = 44||.06|
|PCVa (%)||50 (28–70); n = 21||43.25(29–63); n = 44||<.05|
|WBCa (μL)||26700 (7900–45900); n = 21||17700 (4200–74400); n = 44||<.05|
|RBCa (106/μL)||10.74 (5.9–14.3); n = 21||8.0 (4.8–18.2); n = 44||<.05|
|SUNa (mg/dL)||193.5 (36.3–313.3); n = 25||238.2(93–400.6); n = 45||.14|
|Creatininea (mg/dL)||6.6 (2.0–29.8); n = 25||9.5 (2.0–33.5); n = 45||.07|
|SUN/Creatinine ratioa||23.1 (9–93.2); n = 25||21.7 (5.7–107.3); n = 45||.97|
|ASTa (U/L)||39 (17.8–271); n = 21||53 (12–249); n = 41||.47|
|ALTa (U/L)||56.1 (18.2–287.9); n = 21||56 (13.7–581.5); n = 41||.9|
|LDHa (IU/L)||439 (152–1774); n = 21||252.5 (51–1153); n = 42||<.05|
|CKa (mg/dL)||559 (121–6354); n = 21||368 (48–4477); n = 42||.26|
|Albuminb (g/dL)||4.2 (0.8); n = 21||3.6 (0.7); n = 42||<.005|
|A/G ratiob||1.26 (0.3); n = 21||1.05 (0.3); n = 42||<.05|
|Calciumb (mg/dL)||9.2 (1.0); n = 10||9.2 (1.8); n = 27||.82|
|ALPa||36 (13–190); n = 21||27 (11–219); n = 42||.651|
|Phosphorusb (mg/dL)||14 (5.9); n = 10||15 (5.7); n = 28||.504|
|Chloridea (mEq/L)||101.2 (82–160); n = 15||114 (88–156); n = 25||.09|
|Glucosea (g/dL)||198 (120–309); n = 20||147 (32–375); n = 43||<.005|
|Total proteinb (g/dL)||7.6 (1.1); n = 22||7.3 (1.3); n = 42||.28|
|Sodiuma||144.1 (130–198); n = 15||150 (116–199); n = 25||.14|
|Potassiuma (mEq/L)||4.2 (3.1–9.9); n = 14||4.6 (1.5–10); n = 24||.77|
|DSUN||159.6 (−79.6–273); n = 22||52.9 (−30.7–203); n = 16||<.05|
|DCreatinine||5.2 (−1.8–7.5); n = 22||1.3 (−1.5–27.5); n = 15||<.05|
After univariate analysis, higher levels of BT, PCV, RBC, glucose, and albumin, as well as a higher A/G ratio, were negatively associated with death in cats with AKI, while age was positively associated (all P < .05). Additionally, both DSUN and DCr had significantly predictive ability in terms of future survival, as it was found that a greater decrease in SUN and creatinine concentrations over 3 days were related to a reduced odds ratio of death due to AKI (Table 3).
|Variable||Odds Ratio||95% CI||P Valuea|
Further analysis of the data by multiple logistic regression with backward selection was then carried out. Factors of BT, albumin, and LDH were selected as the most appropriate variables to incorporate in the model to predict death due to AKI (Table 4).
|Variable||aOR (95% CI)||P Value|
|Body temperature||0.509 (0.296–0.875)||.015|
Based on this equation, the log odds was calculated for each study individual to generate a summarized prognostic index for cats with AKI. A higher log odds value indicates a higher probability of death. Using this index to determine risk of future death due to AKI, the AuROC was 0.860 ± 0.026. It was further determined that using this index, a cut-off value of 0.822 had a sensitivity of 77% and a specificity of 90% when determining future death among cats with AKI (Fig 1).
In this study, a summary prognostic index for predicting survival among cats with AKI was obtained by multiple logistic regression analysis. This method has been widely applied to study the prognosis of various different diseases. This study showed that lower BT, concentration of serum albumin or LDH activity are useful predictors of death in cats with AKI. The values for these indices, when the survival and nonsurvival groups were considered, are also significantly different. Furthermore, when BT, serum albumin, and LDH were considered concurrently, there was a good predictive value for the prognosis of cats with AKI. Using this information, it was possible to develop a summary prognostic index, and a cut-off value of 0.822 was demonstrated to have high sensitivity (77%) and specificity (90%) when predicting future risk of death among cats with AKI.
Lower BT values have been reported to be related to chronic kidney disease, but have been seldom considered as a prognosis factor for AKI. Nevertheless, in the present study, BT had a significant predictive value among cats with AKI. It is well known that the kidney plays a role in thermogenesis, and that nephrectomized animals have lower core temperature; in these circumstances, it is quite probable that hypothermia might occur due to renal disease. Additionally, in an experimental study, hypothermia has been found to induce ultrastructural changes in renal tubular cells and these changes were similar to acute tubular necrosis. Therefore, a low body temperature might result from renal injury and may then lead to further renal functional deficiencies.
In a previous study, a decrease in serum albumin increased the odds of death in the cats with acute kidney injury. However, in our study, the concentration of albumin and the A/G ratio in the survival and nonsurvival groups were not significantly lower than the normal reference value, indicating that hyperalbuminemia might be a result of dehydration among the cats with AKI. Dehydration hemodynamically decreases glomerular filtration rate and this in turn results in prerenal azotemia, and such injury can be resolved by correcting the hydration status. Nevertheless, prolonged dehydration can cause renal tubular necrosis and exacerbate the kidney injury. Although an elevation of the SUN/creatinine ratio can be used to evaluate prerenal azotemia, there was no significant difference in SUN/creatinine ratio between surviving and nonsurviving cats with AKI in our study. However, the case clinical information in this study was derived from historical medical records and therefore the magnitude or the duration of any dehydration, as well as any prerenal factors that might affect each cat, could not be accurately accessed.
In this study, the mean and median LDH activity for the survival and nonsurvival groups were significantly higher than the normal reference value. Notably, previous studies have suggested that a high activity of serum LDH might reflect renal tubular damage or a renal infarct. LDH was a prognostic factor of a good outcome in this study. As LDH comes from different organs and contained several types of isoenzymes, the total serum LDH activity might not be specific to renal disease. Nevertheless, the reason why cats suffering AKI having a high serum LDH are more likely to survive requires further investigation.
The case fatality rate of cats with AKI was 64% in our study, which is higher than the previous reports for humans, cats (47%), and dogs.[4, 5] Nevertheless, such a high case fatality rate means that it is important that there is an early identification of prognostic factors when cats suffer from AKI. Six cats with auria died within 3 days of hospitalization in our study (data not shown). The most common causes of AKI in cats have been shown to be related nephrotoxicoses, possibly through exposure of antibiotics,[17, 19] ingestion of lily,[18, 20] nonsteroid anti-inflammatory medicine(NSAID) and ethylene glycol. Unfortunately, most of the causes remain unknown in this study, due to the incomplete history taking or owner's ignorance. Nevertheless, only 2 cats were with NSAID treatment and 1 cat received antibiotics according to the clinical history. Future studies need to take the information into consideration for constructing a more applicable model of prognosis.
Different treatments might result in diversity of clinical outcomes. In this study, all the cats were treated with fluid therapy intravenously to correct the hydration status, electrolytes and acid-base balance. Additionally, the clinical assessments of mucous membrane capillary refill time, heart and respiratory rate, and body weight were utilized to adjust the fluid rate. Diuretics were used during auria or oliguria status. Oliguria was defined as urine production less than 6.5 mL/kg/day and the urine output was approximately measured by litter box weight. Nevertheless, quantitative urine output is an important risk factor in relation to acute kidney injury in both human and veterinary medicines. In the current standard managements of AKI, monitoring fluid ins and outs by a urinary catheter and serial monitoring of central venous pressure (CVP) are suggested to facilitate safety of fluid administration and lower the fertility rate of AKI in cats. Unfortunately, the precise urine output and CVP of each cat could not be obtained as these works are not routinely conducted in our hospital. Therefore, their impact on AKI prognosis could not be evaluated. In the future, it is important that the current model needs to take this clinical management factor into account for better prognosis evaluation.
Conventionally, elevation of SUN and creatinine (azotemia) are considered the key prognostic factors of human and canine AKI. In this study, the results suggested that neither SUN nor creatinine is increased when a cat suffers death from AKI, which is consistent with a previous report. However, the above variables should not be neglected because consistently decreased concentrations of SUN and creatinine in 3 days were found to be associated with a lower case fatality rate in this study. The deliberate monitoring of serum SUN and creatinine may therefore be useful when assessing prognosis among cats with AKI.
Geriatric cats had a higher case fatality rate from AKI in this study. Similar results had been reported in dogs and humans. Oxidative stress can result in geriatric arterioles, which in turn might result in a decline in the renal glomerular filtration rate (GFR); subsequent to this, the pathological stress from AKI might cause further deterioration in renal function. Based on this, elderly cats with AKI should be cautiously treated.
Renal failure might induce hyperglycemia via insulin resistance, which can then lead to stress hyperglycemia. Interestingly, in the present study, the blood glucose concentration of the survival cats was higher than that of the nonsurvival ones. The serum glucose concentration increased in parallel with the chance of survival. This might be associated with stress hyperglycemia as a result of struggling by the cats. It can be hypothesized that cats in a better physical condition are able to struggle harder, which might contribute to higher serum glucose, although the exact mechanism needs further study.
This work found that both the survival and nonsurvival groups had elevated leukocyte counts. Notably, the leukocyte count in the survival group was significantly higher than in the nonsurvival group. The increase in leukocytes might be associated with a poor prognosis. Despite its complexity, the activation and suppression of the uremia-related immune response have been proposed to be related to AKI. In the acute-phase response, infection or massive tissue damage can lead to an increase in leukocytes. However, immune mediators, such as cytokines, protease, prostaglandins, and uremic toxin, might interfere with leukocyte production. In such circumstances, it is difficult to interpret the severity of AKI in cats merely examining leukocyte concentrations.
In the present study, PCV, RBC, glucose, and the A/G ratio were significantly related to death due to AKI in cats by univariate analysis. However, after multivariate analysis with backward selection, these factors were excluded and only BT, albumin, and LDH were selected as the best variables to incorporate in a model predicting death due to AKI. These exclusions can be explained by the fact that albumin probably represents the state of dehydration and that the roles of PCV, RBC, glucose, and the A/G ratio would be similar. Therefore, to prevent collinearity in the multivariate model, only albumin needed to be incorporated into the final model.
In conclusion, a method of evaluating the outcome of AKI at an early stage allows the veterinarian to plan appropriate medical management for cats with AKI. This study pinpoints several prognostic factors and has developed a new summary prognostic index that was modeled using the multiple logistic regression approach. Of major importance, SUN and serum creatinine concentration cats with AKI was not included in this index as it was not statistically useful for prognosis. Nevertheless, as different etiologies of AKI are known to have different prognoses, the cats with AKI in this study might not be representative of the cases treated elsewhere. Therefore, the current model would still need to be further evaluated in the other populations. As a whole, our results are a step forward in predicting the outcome of cats with AKI, and the summary index should be a useful tool to veterinarians in the future.
The authors thank Professor M-L Wong (National Chung-Hsing University) for his critical reading and Professor T-H. Hsu (National Chung-Hsing University) for his kind permission to use the data from the Veterinary Medical Teaching Hospital, National Chung Hsing University.
This study was not supported by any grants.
- 26Comparison of plasma/serum urea and creatinine concentrations in the dog: A 5-year retrospective study in a commercial veterinary clinical pathology laboratory. J Vet Med A Physiol Pathol Clin Med 2004;51:119–123., , , et al.