J Vet Intern Med 2009;23:544–551

Per ipheral Nuc leated R ed Blood Ce lls as a P rognosti c I ndicator

i n H e a t s t r ok e i n D og s
I. Aroch, G. Segev, E. Loeb, and Y. Bruchim

Background: Heatstroke in dogs is often fatal and is associated with a high prevalence of secondary complications. Periph-
eral nucleated red blood cells (NRBC) occur in dogs with heatstroke, but their association with complications and the outcome
is unclear.
Hypothesis: Peripheral NRBC are common in dogs with heatstroke and have prognostic significance.
Animals: Forty client-owned dogs with naturally occurring heatstroke.
Methods: Prospective, observational study. Dogs were followed from presentation to discharge or death. Serum biochem-
istry and coagulation tests were performed at presentation. CBC and evaluation of peripheral blood smears were performed at
presentation and every 12 hours. The relative and the absolute NRBC numbers were calculated.
Results: Presence of NRBC was observed in 36/40 (90%) of the dogs at presentation. Median relative and absolute NRBC
were 24 cells/100 leukocytes (range 0–124) and 1.48  103/mL (range 0.0–19.6  103/mL), respectively. Both were significantly
higher in nonsurvivors (22) versus survivors (18) and in dogs with secondary renal failure and DIC versus those without these
complications. Receiver operator curve analysis of relative NRBC at presentation as a predictor of death had an area under
curve of 0.92. A cut-off point of 18 NRBC/100 leukocytes corresponded to a sensitivity and specificity of 91 and 88% for death.
Conclusions and Clinical Importance: Relative and absolute numbers of peripheral NRBC are clinically useful, correlate
with the secondary complications, and are sensitive and specific markers of death in dogs with heatstroke, although they should
never be used as a sole prognostic indicator nor should they replace clinical assessment.
Key words: Acute kidney injury; Disseminated intravascular coagulation; Dog; Metarubricyte; Rubricyte.

eatstroke in dogs is a severe clinical syndrome char-

H acterized by core temperatures 4 411C (105.81F)
with central nervous system (CNS) dysfunction and
Abbreviations:
aNRBC absolute nucleated red blood cells
aPTT activated partial thromboplastin time
tachypnea.1,2 It can result from exposure to a hot and
CI confidence interval
humid environment (classical or environmental heat-
CNS central nervous system
stroke) or from strenuous physical exercise (exertional
DIC disseminated intravascular coagulation
heatstroke). Heatstroke is quite common in dogs and oc- HUVTH Hebrew University Teaching Hospital
curs particularly during the summer months, mainly in NRBC nucleated red blood cells
hot and humid environments.1,2 It is associated with a PT prothrombin time
systemic inflammatory response syndrome (SIRS), lead- rNRBC relative nucleated red blood cells
ing to multiple organ dysfunction and death.3,4 ROC receiver operating characteristics
Clinical heatstroke or heat-related illness in dogs has SD standard deviation
been described in 2 relatively large retrospective studies SIRS systemic inflammatory response syndrome
with an overall mortality of 50 and 64%.1,2 Risk factors
for death include obesity, prolonged (4 90 minutes) time
lag from the heat insult to presentation, hypoglycemia death.1 Additionally, nonsurvivor dogs have lower glucose,
(o 47 mg/dL) on admission, azotemia (creatinine total protein, albumin, and cholesterol concentrations
4 1.5 mg/dL at 24 hours after presentation), and develo- and significantly higher concentrations of creatinine and
pment of disseminated intravascular coagulation (DIC) total bilirubin than survivors. Nonsurvivors also have a
and acute kidney injury.2 Mental status on presentation higher occurrence of ventricular arrhythmias.1
(comatose versus noncomatose) is associated with Because of the high rate of systemic complications and
fatality in heatstroke, treatment is challenging, complex,
and costly, whereas the prognosis is uncertain. Although
From the Koret School of Veterinary Medicine, Hebrew University some of the above-mentioned risk factors can be used to
of Jerusalem, Rehovot, Israel. The patients were treated at the Koret
School of Veterinary Medicine, Hebrew University of Jerusalem, Is-
assess the disease severity and the prognosis at presenta-
rael. Partial results of this study were presented at the 17th Annual tion, others cannot be used at that time and, thus, cannot
ECVIM-CA Congress, Septembr 13–15, 2007, Budapest, Hungary. facilitate prognostic projections. Therefore, a readily
Corresponding author: Itamar Aroch, DVM, DECVIM-CA (In- available, cost-efficient tool that can accurately predict
ternal Medicine), Koret School of Veterinary Medicine, Hebrew the severity of illness and the outcome in dogs with heat-
University of Jerusalem, P.O. Box 12, Rehovot 76100, Israel; stroke and assist in therapeutic planning would be
e-mail: aroch@agri.huji.ac.il.
extremely useful.
Submitted June 15, 2008; Revised January 20, 2009; Accepted
January 21, 2009.
Peripheral nucleated red blood cells (NRBC) are re-
Copyright r 2009 by the American College of Veterinary Internal ported in blood smears of 15/26 dogs (58%) with heat-
Medicine related illness.1 NRBC occur in mildly, moderately, and
10.1111/j.1939-1676.2009.0305.x severely affected animals, with a median count of 2.5 and
 Heatstroke in Dogs
1.0 cells per 100 white blood cells (WBC) in survivors and
nonsurvivors, respectively. Counts in 16 survivors and 8
nonsurvivors were 0–3 and 0–95 cells per 100 WBC, re-
spectively.1 Peripheral NRBC decreased by  66% in 3
dogs, which were sampled sequentially. However, there
was no information when this was evident in relation to
presentation or heat insult.1 NRBC are detected in 13/19
dogs with heatstroke (68%) (median relative NRBC 10,
range 1–67 cells/100 leukocytes).2 The NRBC to
polychromatophils ratio was 4 1 in all cases, indicating
a pathologic process unrelated to erythroid hyperplasia.2,3
The aims of this study were to prospectively investigate
the prevalence, characteristics, and association of periph-
eral NRBC with clinical and clinicopathological data
in naturally occurring heatstroke in dogs and to assess
its usefulness as a marker of secondary complications,
such as acute kidney injury and DIC, and as a prognostic
indicator.
Materials and Methods
Selection of Cases and Collection of Data
This prospective study was approved by the Animal Care and
Use Committee of the Hebrew University Veterinary Teaching Hos-
pital (HUVTH). Dogs diagnosed with heatstroke based on the
following criteria were enrolled prospectively and consecutively into
the study. All dogs were presented to the Emergency Service,
HUVTH, between May 2005 and September 2006, and diagnosed
with heatstroke based on the history and presence of characteristics
clinical signs, which had developed only after exposure to a warm
environment, strenuous activity, or both. These signs included col-
lapse, with or without CNS abnormalities. Collapse was defined by
a history or physical examination findings of an acute loss of ambu-
lation, recumbency, and inability to stand, which had occurred
acutely. Dogs with coexisting medical conditions were excluded.
Data included the signalment, history, clinical signs on admission
and during hospitalization, laboratory test results, hospitalization
time period (from presentation to discharge or death), outcome
(survived to discharge, dead, or euthanized during hospitalization),
and necropsy results. Environmental heatstroke was diagnosed if
the clinical signs occurred after exposure to a hot and humid envi-
ronment with no history of strenuous physical activity, whereas
exertional heatstroke was diagnosed if the history indicated that the
clinical signs appeared after strenuous physical exercise.2 The at-
tending clinicians in charge of cases were blind to the NRBC count
results evaluated in the current study. At presentation, dogs were
assigned a neurological status score based on a 0–4 scale, where 0 5
alert to mild depression, 2 5 moderate to marked depression or de-
lirium, 3 5 stupor, and 4 5 coma. Delirium was defined as a state of
aimless walking, odd responses to stimuli, staggering, and partial
response to the dog’s name. Stupor was defined as a state of near-
unconsciousness of severe diminished consciousness in which the
dog responded only to strong stimuli, such as pinching with a he-
mostat. Coma was defined as a state of complete loss of
consciousness with no response to strong stimuli.
Definition of Secondary Complications
Dogs diagnosed with DIC had thrombocytopenia (platelets
o 150,000/mL, reference range 150,000–500,000/mL) and at least
two of the following: prolongation (4 25%) of prothrombin time
(PT) or activated partial thromboplastin time (aPTT) and clinical or
postmortem signs compatible with DIC, including hematochezia,
hematemesis, hematuria, and diffuse bleeding at any time point
545
during the disease course.2 Dogs were diagnosed with acute kidney
injury if the serum creatinine concentration was 4 2 mg/dL after 24
hours of IV administration of fluids and the exclusion of pre- and
postrenal components of their azotemia, or based on histological
evidence of acute tubular necrosis at necropsy. Aggressive IV ad-
ministration of fluids for 24 hours was assumed to have eliminated
hypovolemia as a contributing prerenal mechanism for increased
serum creatinine concentration. The urinary bladder of all dogs was
routinely catheterized and the urine output was continuously mon-
itored throughout the hospitalization thus excluding any postrenal
mechanism for the increase in serum creatinine concentration.
Laboratory Tests
Blood samples for CBCa and coagulation testsb were collected in
potassium-EDTA and trisodium-citrate tubes, respectively, and an-
alyzed within 30 minutes from collection. CBC, differential WBC,
and NRBC counts and PT and aPTT were performed at presenta-
tion (and before treatment) and q8h–q12h until discharge from the
hospital or until death or euthanasia. Differential WBC counts were
performed manually by counting 100 leukocytes in modified
Wright’s-stained blood smears.c Total and differential NRBC
counts and assessment of blood cell morphology were performed
manually on examination of the same peripheral blood smears.
NRBC were counted as NRBC per 100 WBC (rNRBC) when ob-
served and their absolute number (aNRBC) was calculated based on
the total WBC. The WBC count was corrected accordingly.5
Blood for glucose and creatinine concentration measurement was
collected in sodium fluoride and serum tubes, respectively, and was
analyzedd within 24 hours from collection.
Statistical Analysis
For all continuous parameters, the normality of data distribution
was evaluated by means of the Shapiro–Wilk test. Normally and
nonnormally distributed continuous parameters are reported as
mean  standard deviation (SD) and as median and range, respec-
tively, and were compared between the survivors and the
nonsurvivors (euthanized cases excluded) by means of the Student
t- or Mann–Whitney tests, respectively. Fisher’s exact test was used
to compare categorical variables between the outcome groups. Lo-
gistic regression analysis was performed to assess the relationship of
different variables with the outcome. The association between
aNRBC and rNRBC with the outcome was evaluated in all the
dogs in the study as well as after exclusion of the euthanized animals
by the receiver operating characteristics (ROC) procedure. The
ROC analysis was used to select NRBC cut-off points and their
corresponding sensitivities and specificities for prediction of
the outcome. The area under the ROC curve was calculated by the
trapezoidal rule. Fisher’s exact test was used for the calculation of
the 95% confidence interval (CI95%) for the specificity and sensitiv-
ity. The optimal cut-off point was selected as the point that was
associated with the least number of misclassifications. Spearman’s
rank correlations were used to assess the correlation between con-
tinuous variables. For all tests applied, P o .05 was considered
statistically significant. All calculations were performed using a sta-
tistical software.e
Results
Signalment History and Clinical Signs
Forty dogs met the inclusion criteria, of which 23 and
17 had exertional and environmental heatstroke, respec-
tively. All dogs were admitted from April to October,
which corresponds to the warm season in Israel. The
dogs had a median body weight of 34 kg (range 6–60) and
546 Aroch et al
included the following breeds: mixed-breed, Golden Re-
triever, Labrador Retriever, Belgian Malinois, Boxer,
Chinese Shar-Pei, dog de-Bordeaux, Rottweiler (12, 5, 4,
4, 2, 2, 2, and 2 dogs, respectively), Great Dane, English
Bulldog, Staffordshire Bull-Terrier, German Shepherd,
English CockerSpaniel, Rhodesian Ridgeback, and Aire-
dale Terrier (1 dog each). There was no significant
difference between the proportion of males (25) and fe-
males (15) (P 5 .36). The median time lag from the insult
of hyperthermia to presentation at the HUVTH was 3
hours (range 0.5–24). Twenty-five of 39 dogs (64%) in
which this information was available were cooled by
their owners before presentation at the HUVTH.
The most common clinical signs at presentation in-
cluded collapse (39 dogs, 98%, CI95% 92–100%), shock
(31 dogs, 79%, CI95% 66–93%), and seizures (14 dogs,
37%, CI95% 21–53%). The mental status was assessed in
36 dogs and abnormalities were observed in most of
them, including coma (11 dogs, 31%, CI95% 15–46%),
stupor (14 dogs, 39%, CI95% 22–56%), and delirium (10
dogs, 28%, CI95% 12–43%), whereas only 1 dog (3%,
CI95% 0–8%) presented with mild depression with no ad-
ditional neurological abnormalities. When comatose
(mental status score 4) and no comatose (mental status
1, 2, and 3 combined) dogs were compared, the propor-
tion of nonsurvivors was significantly (P 5 .01) higher in
the comatose group. The median body temperature at
presentation was 39.1 1C (SD 2.1, range 35.5–43.3, CI95%
38.5–39.8 1C). During hospitalization, petechiae or echy-
moses and bloody diarrhea were observed in 25 (63%)
and 19 (48%) dogs, respectively.
Laboratory Results
The case fatality rate and the rapid disease progression
along with technical limitations led to loss of data in
some dogs. The most prevalent abnormalities at presen-
tation included presence of peripheral NRBC (95%,
CI95% 87–100%), increased hematocrit, hemoglobin con-
centration, and red blood cell (RBC) count (75, 78, and
68%, respectively, CI95% 61–89, 64–91, and 52–83%, re-
spectively), thrombocytopenia (54%, CI95% 37–70%),
absolute lymphopenia (38%, CI95% 22–54%), ne-
utropenia (28%, CI95% 13–43%), leukocytosis (25%,
CI95% 11–39%), monocytopenia (25%, CI95%11–39%),
leukopenia (15%, CI95% 3–27%), and neutrophilia
(15%, CI95% 3–27%). The median rNRBC and aNRBC
were 24 per 100 WBC (range 0–124) and 1.48  103/mL
(range 0.0–19.6  103/mL), respectively (Table 1). The
majority of NRBC were metarubricytes (median 86%,
range 50–100% of all NRBC) and the minority was ru-
bricytes (median 14%, range 0–50% of all NRBC).
Earlier erythroid precursors were not detected. The
NRBC to polychromatophil ratio was 4 1 in all smears.
The relative and absolute NRBC as well as the number of
dogs presenting NRBC of all dogs decreased whereas the
percent metarubricytes of total NRBC increased from
presentation to 36 hours after presentation; however, the
number of dogs available for examination at subsequent
time points after presentation was limited, mostly due to
death (Table 2). When considering only data from dogs
available at both presentation and 24 hours after presen-
tation, the findings were similar to those of the entire
population. Median rNRBC decreased from 18 to 2 per
100 WBC within the first 24 hours and median aNRBC
decreased from 14.9  103/mL to 2.5  103/mL. Con-
versely, the metarubricyte percentage of total NRBC
increased from 85 to 94%.
Prolongation of the PT and the aPTT at presentation
were documented in 76% (CI95% 61–90%) and 49%
(CI95% 32–66%) of the dogs, respectively. Hypoglycemia
and increased serum creatinine were present in 53%
(CI95% 37–70%) and 49% (CI95% 32–66%) of the dogs,
respectively (Table 1).
Secondary Complications and Outcome
Information of the kidney status was available for 38/
40 dogs. Sixteen (42%) and 23 (58%) dogs were diag-
nosed with acute kidney injury and DIC, respectively.
Eighteen (45%), 19 (48%), and 3 (8%) dogs survived,
died naturally, and were euthanized, respectively. The 3
dogs were euthanized at 2, 24, and 48 hours postpresen-
tation due to financial considerations. Twenty of the 22
nonsurvivors died or were euthanized within 36 hours
from presentation. The median hospitalization time pe-
riods of all dogs, nonsurvivors, and survivors were 24,
12, and 60 hours (all dogs range, 2 hours to 16 days), with
a significant (P 5 .0002) difference between survivors
and nonsurvivors. The death rates among patients with
acute kidney injury and DIC were 81% (13/16) and 82%
(19/23), respectively. The death rate among dogs with
coexisting acute kidney injury and DIC was 87% (13/15).
Clinicopathologic Data in Survivors and Nonsurvivors
At presentation, nonsurvivors had significantly longer
PT (P 5 .01), aPTT (P 5 .0001), higher counts of abso-
lute lymphocytes (P 5 .018), rNRBC (P o .0001, Fig 1),
aNRBC (P 5 .0001, Fig 1), percent rubricytes of all
NRBC (P o .001), absolute metarubricytes and rubric-
ytes (P 5 .0001 and .0001, respectively), lower platelets
(P 5 .023), higher creatinine (P 5 .01), and lower glucose
(P 5 .006) concentrations compared with survivors (Ta-
ble 3). The median rNRBC of survivors and
nonsurvivors on follow-up blood counts were available
only for 15 dogs (9 survivors and 6 nonsurvivors) at 12
hours and for 13 dogs (8 survivors and 5 nonsurvivors) at
24 hours. The median rNRBC at 12 hours was 6 (range
0–17) and 10 (range 2–15.5) in survivors and nonsurvi-
vors, respectively, and at 24 hours was 0.5 (range 0–12)
and 2.5 (range 1–12), respectively. At these two time
points, however, the rNRBC was not statistically differ-
ent between the two outcome groups.
Correlations of NRBC with Clinicopathologic
Measures
There were significant correlations between the relative
rNRBC and PT (r 5 0.45, P 5 .007), aPTT, (r 5 0.75,
P o .0001), WBC (r 5 0.36, P 5 .026), absolute lym-
phocyte count (r 5 0.46, P 5 .004), and glucose concen-
tration (r 5 0.59, P 5 .02). There were significant
 Heatstroke in Dogs 547

Table 1. Hematology and coagulation measures and serum glucose and creatinine concentrations in dogs with heat-
stroke at presentation.
Measure n Median Range % o RI (n) % 4 RI (n) Reference Interval
Red blood cells (106/mm3) 40 8.7 5.5–13.0 0 (0) 68 (27) 5.5–8.0
Hemoglobin (g/dL) 40 19.8 5.0–24.0 0 (0) 78 (31) 12.0–17.5
Hematocrit (%) 40 59.3 34.2–80.8 3 (1) 75 (30) 37–50
Mean corpuscular volume (fL) 40 66.0 61.0–72.0 0 (0) 0 (0) 60–77
Mean corpuscular hemoglobin (pg) 40 22.2 18.2–25.2 5 (2) 0 (0) 19.5–24.5
MCHC (g/dL) 40 33.2 27.0–36.6 13 (5) 5 (2) 32.0–36.0
RDW (%) 40 17.5 16.5–21.4 0 (0) 5 (2) 14.0–19.0
White blood cells (103/mm3) 40 10.80 1.31–45.40 15 (6) 25 (10) 6.00–17.00
Corrected white blood cellsa (103/mm3) 39 8.5 1.2–30.3 13 (5) 23 (9) 6.00–17.00
Platelets (103/mm3) 39 137 0–537 54 (21) 3 (1) 150–500
Segmented neutrophils (103/mm3) 39 6.10 0.83–18.13 28 (11) 15 (6) 3.00–11.50
Band neutrophils (103/mm3) 39 0.00 0.00–10.93 NA 8 (3) 0.00–0.30
Lymphocytes (103/mm3) 39 1.67 0.10–6.45 38 (15) 5 (2) 1.00–4.80
Eosinophils (103/mm3) 39 0.29 0.00–1.45 NA 8 (3) 0.00–1.00
Monocytes (103/mm3) 39 0.35 0.00–1.87 25 (10) 8 (3) 0.10–1.36
Basophils (103/mm3) 39 0.00 0.00–0.61 NA 1 (3) Rare
rNRBC (cells/100 white blood cells) 39 23.5 0.0–124.0 NA 90 (36) None
Absolute NRBC (103/mm3) 39 1.48 0.00–19.61 NA 90 (36) None
Metarubricytes (% of total NRBC) 36 86 50–100 NA NA NA
Rubricytes (% of total NRBC) 36 14 0–50 NA NA NA
Prothrombin time (seconds) 37 10 5.1–60.0 3 (1) 76 (28) 6.0–8.5
aPTT (seconds) 37 19.3 9.9–70.0 5 (2) 49 (18) 11.5–19.5
Glucose (mg/dL) 36 45 12–266 53 (19) 8 (3) 70–110
Creatinine (mg/dL) 37 1.5 0.5–3.7 0 (0) 49 (18) 0.50–1.50
a
Total nucleated cells minus nucleated red blood cells.
aPTT, activated partial thromboplastin time; MCHC, mean corpuscular hemoglobin concentration; NA, not applicable; NRBC, nucleated
red blood cells; RDW, red blood cell distribution width; RI, reference interval; rNRBC, relative nucleated red blood cells (cells per 100
leukocytes).

correlations between aNRBC and WBC (r 5 0.62,
P o .001) and lymphocyte count (r 5 0.51, P 5 .012),
but not with neutrophil and monocyte numbers.
NRBC at Presentation as a Predictor of Secondary
Complications and Death
Dogs with DIC had significantly (P o .001) higher
rNRBC (median 55, range 7–124 per 100 WBC) com-
pared with those without DIC (median 2, range 0–41
cells per 100 WBC). Dogs with acute kidney injury had
significantly (P 5 .006) higher rNRBC (median 48, range
0–76 cells per 100 WBC) compared with dogs without
acute kidney injury (median 9, range 0–76 cells per 100
WBC). Receiver operator curve analysis of rNRBC (cells
per 100 WBC) at presentation as a predictor of DIC had
an area under the curve (AUC) of 0.94 (CI95% 0.86–1.00).
A cut-off point of 13 rNRBC per 100 WBC corresponded
to a sensitivity and specificity of 86% (CI95% 65–97%)
and 87% (CI95% 60–98%), respectively. In a similar anal-
ysis, performed to assess rNRBC as a predictor of acute
kidney injury, the area under the ROC curve was 0.76
(CI95% 59–93%) and a cut-off point of 17 rNRBC per
100 WBC yielded a sensitivity and specificity of 87%
(CI95% 60–98%) and 67% (CI95% 43–85%), respectively.
Relative NRBC was found as a sensitive and specific
marker of death. Receiver operator curve analysis of
rNRBC (cells per 100 WBC) at presentation as a predic-
tor of death yielded an AUC of 0.90 (CI95% 0.80–1.00). A
cut-off point of 18 rNRBC corresponded to a sensitivity
and specificity of 91% (CI95% 70–99%) and 88% (CI95%
64–99%), respectively. An rNRBC cut-off point of 5
cells/100 leukocytes corresponded to a sensitivity and
specificity of 100% (CI95% 84–100%) and 59% (CI95%
33–82%), respectively. An rNRBC cut-off point of 38
cells/100 leukocytes corresponded to a sensitivity and

Table 2. Changes in nucleated red blood cells during hospitalization in dogs with heatstroke.
Time Postpresenta- Number (%) of Dogs with Median (range) rNRBC Median (range) Absolute Median (range) % Meta-
tion (hours) NRBC of All Dogs (cells per 100 WBC) NRBC ( 103/mm3) rubricytes of Total NRBC
0 36/38 (95) 24 (0–124) 1.48 (0.00–19.61) 86 (50–100)
12 12/15 (80) 6 (0–17) 0.00 (0.00–2.40) 96 (80–100)
24 7/13 (64) 1 (0–12) 0.00 (0.00–1.60) 100 (75–100)
36 2/4 (50) 0 (0–2) 0.06 (0.00–0.13) 100 (100–100)

NRBC, nucleated red blood cells; rNRBC, relative nucleated red blood cells (cells/100 leukocytes); WBC, white blood cells.
548 Aroch et al

under the ROC for relative and absolute NRBC in-

creased further to 0.92 (CI95% 0.82–1.00) (Fig 2) and
0.89 (CI95% 0.77–1.00), respectively. Optimal cut-off
points with corresponding sensitivities and specificities
for these analyses were 18 cells per 100 leukocytes, 95%
(CI95% 74–100%), and 88% (CI95% 64–99%) for
rNRBC, respectively, and for 800 cells/mL, were 95%
(CI95% 74–100%) and 77%, (CI95% 50–93%), respec-
tively, for aNRBC.

Fig 1. Relative nucleated red blood cells (NRBC) (cells per 100
leukocytes) at presentation and absolute NRBC in 17 survivor and
21 nonsurvivor dogs with heatstroke.
specificity of 67% (CI95% 43–87%) and 94% (CI95% 71–
100%), respectively. When the aNRBC were used instead
of the rNRBC, the area under the ROC was 0.86 (CI95%
0.74–0.98) and the optimal cut-off point of 800 cells/mL
corresponded to 91% (CI95% 70–99%) sensitivity and
77% (CI95% 50–93%) specificity. When the analyses were
repeated with the 3 euthanized dogs excluded, the area
Discussion
The results of the present prospective study confirm
that the presence of peripheral NRBC is common in dogs
with heatstroke and are in agreement with previous ob-
servations.1,2 In fact, presence of NRBC was the most
common hematological abnormality at presentation in
dogs in this study. Most NRBCs were metarubricytes,
the minority was rubricytes, and earlier erythroid precur-
sors were absent. As the recent history of dogs with
heatstroke is sometimes incomplete, and dogs are often
presented with normal rectal temperatures and even
hypothermia, presence of peripheral NRBC in a col-
lapsed, tachypneic, and neurologically abnormal dog
should thus raise suspicion of presence of heatstroke.

Table 3. Selected laboratory measures at presentation and the hospitalization period in survivor and nonsurvivor
dogs with heatstroke.
Survivors Nonsurvivors

Measures Median (range) % 4 RI % o RI Median (range) % 4 RI % o RI RI P-Value

Prothrombin time (seconds) 10 58.8 5.9 15.1 90 0 6–8.5 .01
(5.1–19.6) 7.3–60.0
aPTT (seconds) 17 11.8 11.8 29.9 80 0 11.5–19.5 o .001
(9.9–34.0) (15.0–70.0)
Platelets ( 109/L) 217 6.3 75 108 0 100 150–500 .023
(0–573) (0–377)
Lymphocytes ( 103/L) 0.9 0 75 2.2 9.5 23.8 1–4.8 .018
(0.0–3.1) (0.0–6.4)
Relative NRBCa (%) 2 88.2 NA 50 100 NA 0 o .001
(0–87) (7–124)
Absolute NRBC ( 103/L) 0.2 88.2 NA 4.6 100 NA 0 o .001
(0.0–11.2) (0.1–19.6)
Metarubricyesb (%) 85 NA NA 100 NA NA NA o .004
(0–100) (50–100)
Absolute metarubricytes ( 103/L) 0.2 88.2 NA 3 95.2 NA 0 o .001
(0.0–10.0) (0.0–16.8)
Rubricytesc (%) 0 41.2 NA 7 NA NA NA o .001
(0–5) (0–21)
Absolute rubricytes ( 103/L) 0 41.2 NA 0.53 81 NA 0 o .001
(0.0–1.2) (0.0–2.8)
Creatinine (mg/dL) 1.1 20 0 1.8 70 0 0.5–1.5 .014
(0.5–2.4) (0.5–3.7)
Glucose (mg/dL) 90 27.3 27.3 30 0 93.9 70–110 .005
(12–266) (10–88)
Hospitalization period (days) 2.5 NA NA 0.6 NA NA NA o .001
(1–14) (0.1–16)
a
Nucleated red blood cells per 100 leukocytes.
b
% Metarubricytes of all nucleated red blood cells.
c
% Rubricytes of all nucleated red blood cells.
aPTT, activated partial thromboplastin time; NA, not applicable; NRBC, nucleated red blood cells; RI, reference interval.
 Heatstroke in Dogs
Fig 2. Receiver operator curve of relative nucleated red blood cells
(cells per 100 leukocytes) at presentation and survival (3 euthanized
dogs excluded) in dogs with heatstroke. AUC, area under the curve.
The presence of peripheral NRBC was not associated
with anemia and erythroid hyperplasia. At presentation,
most of the dogs had increased hematocrit, hemoglobin
concentration, and RBC count. Furthermore, the NRBC
to polychromatophil ratio was 4 1 in all dogs, suggest-
ing a disease process unrelated to erythroid hyperplasia,3
as reported in a previous retrospective study.2 Finally,
the acute onset and rapid progression of the disease in the
dogs excludes a bone marrow reaction with increased
erythropoiesis. Thus, the high prevalence of peripheral
NRBC in dogs with heatstroke should be considered to
be a direct or an indirect result of hyperthermia. Periph-
eral NRBC were also observed in human patients after
thermal injuries and were most commonly present in pa-
tients with the largest burns.6
Appearance of peripheral NRBC in human heatstroke
has never been described; however, in critically ill human
patients, appearance of NRBCs in the peripheral blood
has been associated with a variety of severe diseases and
poor prognosis.7–12
The high prevalence of peripheral NRBC at presenta-
tion in the dogs in the present study made it possible to
assess their usefulness as a prognostic marker for predic-
tion of secondary complications and outcome. At
presentation, significantly higher rNRBC and aNRBC
were observed in nonsurvivors compared with survivors.
It is clear from the results that NRBC numbers, whether
relative or absolute, were predictors of death in dogs with
heatstroke, as reflected by the high sensitivity and spec-
ificity. The presence of NRBC at presentation probably
represents the severity of the thermal injury to the bone
marrow and the latter probably correlates with the se-
verity of the thermal injury of other organs and
subsequent certain secondary complications, such as
acute kidney injury and DIC as well. Thus, presence and
magnitude of NRBC can be used as a simple, cost-effec-
tive, and readily available marker of overall organ injury
in dogs with thermal injury. This is demonstrated by the
finding that the higher the peripheral NRBC number at
549
presentation, the higher was the probability that the dis-
ease has culminated in secondary complications such as
DIC, acute kidney injury, and death. This assumption is
supported by the significantly higher rNRBC at presen-
tation in dogs diagnosed with DIC and acute kidney
injury compared with those that had no such complica-
tions and by the fact that rNRBC was also a highly
sensitive and specific predictor of DIC and death. In ad-
dition, there were highly significant positive correlations
between rNRBC and PT as well as aPTT at presentation.
Both DIC and acute kidney injury have been frequently
previously reported in people and dogs with heatstroke
and both have been found as risk factors for mortality in
dogs.1,2,4,13,14
The presence and magnitude of NRBC should never
be used as a sole prognostic indicator in dogs with heat-
stroke, nor should they replace proper clinical
assessment, although both were found to be highly sen-
sitive and specific markers of the outcome in this
syndrome. The different cut-off points provided allow
flexibility for clinicians, based on their preference, to ei-
ther maximize sensitivity or specificity through
individualized selection of cut-off points.
Although both rNRBC and aNRBC were highly sen-
sitive and specific predictors of the outcome, there is no
obvious explanation for the fact that rNRBC was a bet-
ter one. Probably, the thermal insult and consequent
bone marrow lesions were expressed more accurately in
the rNRBC. Although rNRBC had a positive correlation
with WBC, this correlation was weak, whereas aNRBC
had a stronger correlation with the WBC and lympho-
cyte number. It seems that the aNRBC, which is
calculated based on the WBC, is more influenced by
changes in the WBC. Consequently, a similar thermal in-
sult leading to a similar release of NRBC from the bone
marrow will result in a lower aNRBC in leukopenic dogs
compared with dogs with normal or increased WBC.
Thus, rNRBC probably provides a better reflection of
the lesion in dogs with leukopenia and serve as a better
predictor of the outcome.
Another interesting observation of this study was that
the proportion of metarubricytes of the total NRBC, the
mature erythroid NRBC precursors, was significantly
higher in survivors compared with nonsurvivors,
whereas the proportion of rubricytes was higher in the
nonsurvivor group. Most likely, the hyperthermia in
nonsurvivors was more severe compared with survivors,
resulting in more severe bone marrow lesions in the for-
mer group with subsequent release of earlier erythroid
precursors. Thus, it is possible that the severity of the
hyperthermia-induced bone marrow lesion that leads to
release of NRBCs is positively associated not only with
the numbers of peripheral NRBCs but also with their
character: the more severe the injury, the higher is the
proportion of rubricytes and lower is the proportion of
metarubricytes of the total peripheral NRBCs popula-
tion. This hypothesis gains further support from the
gradual decrease in NRBC numbers and increase in the
proportion of metarubricytes during hospitalization and
healing. If this is true, prediction of the outcome and sec-
ondary complications in dogs with heatstroke can be
550 Aroch et al
improved by assessing the character of peripheral
NRBCs in addition to their number. Further larger stud-
ies are needed to assess whether this observation is indeed
clinically useful.
In a previous study of heat-related illness, the periph-
eral NRBC decreased by 66% within 12 hours from
presentation; however, only 3 dogs had consecutive
CBC evaluations.1 Similarly, in the present study, the
median rNRBC markedly decreased from presentation
to 12 and 24 hours after presentation (24, 6, and 1 cells/
100 leukocytes, respectively). However; NRBC were still
present in 2 of 4 dogs at 36 hours after presentation.
Nonsurvivors consistently had a higher median rNRBC
compared with survivors at all time points postpresenta-
tion; however, this difference did not reach statistical
significance. Future studies with a higher number of
animals are warranted to assess the clinical usefulness
of NRBC during hospitalization for prediction of the
outcome.
Hypoglycemia has been previously recorded in associ-
ation with death in dogs with heatstroke, as observed in
the present study as well, and, in the current study, was
significantly, although moderately, correlated with
rNRBC, which was a good predictor of death. It was
also significantly correlated with prolongation of the PT
and aPTT, which are potential markers of DIC.
Direct renal thermal injury, hypoxia, hypovolemia
with consequent reduced glomerular filtration, coagulative
necrosis due to DIC and microthrombosis, endotoxemia,
and local and systemic release of cytokines and vasoac-
tive mediators during SIRS probably contributed to the
high prevalence of acute kidney injury in dogs with
thermal injury, as has been previously suggested.15 In
addition, as serious muscle damage is extremely common
in dogs with heatstroke, it is possible that rhabdomyoly-
sis with subsequent myoglobinemia and myoglobinuria
could have also contributed to the development of acute
kidney injury, as has been reported in people with heat-
stroke.2,15 This was manifested by an increased creatinine
concentration at presentation in half of the dogs. Because
peripheral NRBC were found to be significantly higher in
dogs with acute kidney injury, presence of high numbers
of peripheral NRBC at presentation in dogs with heat-
stroke patients should alarm clinicians to the presence of
acute kidney injury, even in the absence of increased se-
rum creatinine concentration. This early identification
should prompt aggressive treatment to support kidney
function and minimize further renal damage.
This study has several limitations; first, the definition
of heatstroke in dogs is currently roughly based on hu-
man medicine criteria, with modification of the core
temperature cut-off point. We have used this definition
in our previous study of dogs with heatstroke2; however,
in dogs, there is no clear cut line between heat-related ill-
ness and heatstroke, as the neurological abnormalities
present in the latter condition could be mild in early dis-
ease stages and, thus, may be missed or hard to interpret.
Thus, application of the present results in mild forms of
heat-related illness in dogs should not be carried out
without caution. Second, the number of available ani-
mals, especially during hospitalization, was limited due
to the high mortality and discharge of animals from the
hospital. This limited the statistical power of the com-
parisons of NRBC numbers and character with the
outcome at different time points after presentation. Ad-
ditionally, despite the efforts made to follow the study
protocol, certain samples were missed and led to loss of
data, which could potentially introduce bias to the statis-
tical analyses. Third, there may be a bias in the diagnosis
of secondary complications, such as DIC and acute kid-
ney injury in the study, as these were diagnosed based on
both antemortem clinicopathological measures as well as
postmortem findings. Possibly, the incidence of both of
these secondary complications may have been underesti-
mated in the survivors due to the low sensitivity of the
laboratory tests. On the other hand, inclusion of diag-
noses based on postmortem results may have led to an
erroneously high incidence of secondary complications in
nonsurvivors compared with survivors. Fourth, the an-
temortem diagnosis of DIC could have been improved by
more advanced tests that were not included in this study,
such as antithrombin, d-dimer, fibrin degradation prod-
ucts, and protein C. Finally, as the number of animals
that had no peripheral NRBC at presentation in the cur-
rent study is small, any conclusions regarding their
absence can only be a speculation.
In conclusion, this study shows that presence of pe-
ripheral NRBCs is a common phenomenon in dog with
heatstroke and can be useful in the diagnosis of suspected
cases of heatstroke in dogs. Both the rNRBC and the
aNRBC had a high clinical correlation with secondary
complications and were highly sensitive and specific pre-
dictors of the outcome. Examination of blood smears of
dogs with heatstroke is simple, time and cost-effective, as
well as readily available diagnostic tool and, thus, is very
useful in clinical practice.
Footnotes
a
Abacus or Arcus; Diatron, Wien, Austria
b
KC 1A micro, Lemgo, Amelung, Germany, or ACL 200, Instru-
mentation Laboratory, Milan, Italy
c
Hema-Tek 2000 Slide Stainer, model 4488B; Bayer Corporation,
Elkhart, IN, Stain: Hematek stainpack; Modified Wright’s Stain
d
Cobas-Mira; Roche, Rottkreutz, Switzerland
e
SPSS 14.0 for Windows; SPSS Inc, Chicago, IL
References
1. Drobatz KJ, Macintire DK. Heat-induced illness in dogs: 42
cases (1976–1993). J Am Vet Med Assoc 1996;209:1894–1899.
2. Bruchim Y, Klement E, Saragusty J, et al. Heat stroke in dogs:
A retrospective study of 54 cases (1999–2004) and analysis of risk
factors for death. J Vet Intern Med 2006;20:38–46.
3. Sodikoff CH. Laboratory Profiles of Small Animal Diseases.
A Guide to Laboratory Diagnosis, 3rd ed. St. Louis, MO: Mosby;
2001:104.
4. Bouchama A, Knochel JP. Heat stroke. New Eng J Med
2002;346:1978–1988.
 Heatstroke in Dogs
5. Tvedten H. Referral and in-office laboratories. In: Willard
MD, Tvedten H, Turnwald GH, eds. Small Animal Clinical Diag-
nosis by Laboratory Methods, 2nd ed. Philadelphia, PA: WB
Saunders; 1994:17.
6. Andes WA. Normoblastemia after thermal injury. Am J Surg
1976;131:725–726.
7. Stachon A, Holland-Letz T, Krieg M. High in-hospital mor-
tality of intensive care patients with nucleated red blood cells in
blood. Clin Chem Lab Med 2004;42:933–938.
8. Stachon A, Segbers E, Holland-Letz T, et al. Nucleated red
blood cells in the blood of medical intensive care patients indicate
increased mortality risk: A prospective cohort study. Crit Care
2007;11:R62.
9. Stachon A, Sonderman N, Imohl M, et al. Nucleated red
blood cells indicate high risk of in-hospital mortality. J Lab Clin
Med 2002;140:407–412.
10. Stachon A, Becker A, Kempf R, et al. Re-evaluation of es-
tablished risk scores by measurement of nucleated red blood cells in
551
blood of surgical intensive care patients. J Trauma 2008;65:666–
672.
11. Lehnhardt M, Katzy Y, Langer S, et al. Prognostic signifi-
cance of erythroblasts in burns. Plast Reconstr Surg 2005;115:120–
127.
12. Stachon A, Hollan-Letz T, Kempf R, et al. Poor prognosis
indicated by nucleated red blood cells in peripheral blood is not as-
sociated with organ failure of the liver or kidney. Clin Chem Lab
Med 2006;44:955–961.
13. Lin FY, Wang JY, Chou TC, et al. Vasoactive mediators and
renal haemodynamics in exertional heat stroke complicated by
acute renal failure. Q J Med 2003;96:193–201.
14. Lu KC, Wang JY, Lin SH, et al. Role of circulating
cytokines and chemokines in exertional heatstroke. Crit Care Med
2004;32:399–403.
15. Tan W, Herzlich BC, Funaro R, et al. Rhabdomyolysis and
myoglobinuric acute renal failure associated with classic heatstroke.
South Med J 1995;88:1065–1068.