Study population
The study population consisted of 61 males (48.0%) and 66 females (52.0%), aged 6 months to 11 years, with a median (IQR) of 4 (2–4) years. The dogs weighed between 1.0 and 7.0 kg with a median (IQR) of 2.5 (2.0–3.5) kg. Age did not differ significantly between sexes but males were significantly heavier (median 3.0 kg, IQR 2.3–3.7 kg) than females (median 2.3 kg, IQR 1.8–3.1 kg) (p=0.003).
As shown by MRI, the lesions were virtually equally often focal (71/126 dogs; 56.4%) and multifocal (55/71 dogs; 43.6%), and usually involved both sides of the brain (90/126 dogs; 71.4%). Of focal lesions, most were located in the forebrain (65/71; 91.6%), in approximately two thirds of cases bilaterally (43/65 dogs; 66.1%). Most multifocal lesions were also bilateral (44/55 dogs; 80.0%) (Table S2).
CSF analysis was performed in 92/127 dogs (72.4%) and inflammation was detected in 81/92 dogs (88.0%, 95% CI: 79.8%–93.2%). Brain biopsy was performed in 7/127 dogs (5.5%) and in all cases it confirmed MUO diagnosis. Results of tests for infectious diseases were available for 75/127 (59.1%) dogs and all were negative.
Treatment was applied in 121/127 dogs (95.3%). Six untreated dogs died within 2 days of disease onset (four were euthanatized and two died during MRI examination). All 121 treated dogs received prednisolone – 37/121 dogs (30.6%) as a sole therapeutic agent and the remaining 84/121 dogs as a part of 6 different treatment protocols based on various combinations of azathioprine, cytosine arabinoside, lomustine, cyclosporine, mycophenolate, and radiotherapy, of which the most common was the protocol including prednisolone and azathioprine (32/121 dogs; 26.4%) (Table S3).
Neurological signs resolved in 85/127 dogs (66.9%, 95% CI: 58.4%–74.5%), however, complete remission was observed in 45/127 dogs (35.4%, 95% CI: 27.7%–44.1%). During the observation period, death occurred in 53/127 dogs (41.7%, 95% CI: 33.5%–50.4%). The overall survival time (time from the onset of MUO to death), calculated for 53 dogs whose exact time of death was known ranged from <1 to 2257 days (6.2 years) with a median (IQR) of 71 (19–290) days. Two MUO-unrelated deaths occurred during MRI examination because of complications of general anesthesia rather than MUO itself. In the observation period, 74 dogs (58.3%) were lost to follow up. The follow-up period for these dogs lasted from 14 to 3538 days (9.7 years) with a median (IQR) of 493 (335–1006) days.
Neurological assessment scales
The dogs with MUO obtained from 5 and 18 points (median 16 points, IQR 15–17 points) in the MGCS, from 0 and 12 points (median 5 points, IQR 3–6 points) in the NDS1, and from 0 to 18 points (median 6 points, IQR 3–9 points) in the NDS2 (Fig. S1). The scores obtained by the dog on the three scales were significantly mutually correlated: scores on the NDS1 and NDS2 were strongly positively correlated with each other (Rs =0.87, p<0.001) but strongly negatively correlated with scores on the MGCS (Rs = -0.76, p<0.001 and Rs = -0.70, p<0.001, respectively) (Fig. S2).
Prognostic factors
MUO-specific survival analysis
Median MSS of dogs with MUO when estimated based on all dogs (including censored) was 2257 days (6.2 years) with the 1st quartile (25% of dogs dead) of 162 days (Fig. 1). Most of deaths (46/53; 86.8%) occurred in the first year of observation (Fig. 2).
In the univariable analysis, neurological assessment scales were significantly associated with the risk of MUO-related death – MGCS negatively (HR 0.82, 95% CI: 0.74–0.91; p<0.001) whereas NDSs positively (NDS1: HR 1.30, 95% CI: 1.14–1.49; p<0.001 and NDS2: HR 1.19, 95% CI: 1.09–1.30; p<0.001). Moreover, three other variables proved to be negatively associated with the risk of MUO-related death: body weight (HR 0.69, 95% CI: 0.51–0.95; p=0.020), unilateral lesions in MRI (HR 0.36, 95% CI: 0.17–0.77; p=0.008), and resolution of neurological signs – partial (HR 0.17, 95% CI: 0.08–0.33; p<0.001) or complete (HR 0.12, 95% CI: 0.06–0.25; p<0.001). Age on admission, sex, brain region affected in MRI, and treatment protocol were not significantly associated with MSS (Table S4).
In the multivariable analysis, controlled for the treatment protocol, body weight, and spread of lesions in MRI, two NDSs remained to be significantly positively associated with the risk of MUO-related death (NDS1: HR 1.19, 95% CI: 1.03–1.39; p=0.023 and NDS2: HR 1.13, 95% CI: 1.02–1.25; p=0.016), while modified MGCS did not prove to be significantly associated with the risk of MUO-related death. Furthermore, both the body weight and the spread of lesions were constantly significant in the multivariable analysis (Table S5).
Short-term prognosis – risk of death within first 7 days
During the first 7 days, 11/127 dogs died (8.7%), however, 2/11 dogs accidentally died during MRI, and they were not considered MUO-related deaths. Therefore, the MUO-related case fatality rate was 7.2% (95% CI: 3.8%–13.1%; 9/125 dogs).
None of the demographic or clinical variables were significantly linked to risk of death of MUO within the first 7 days, except for the scores of all three neurological scales (Table S6). Their prognostic accuracy was highest for MGCS (AUROC 96.6%; 95% CI: 92.7%–99.9%), followed by NDS1 (AUROC 95.1%; 95% CI: 86.9%–99.9%) and NDS2 (AUROC 90.6%; 95% CI 79.6%–99.9%), however, the difference between them was not significant (Fig. 3). The optimal cut-off value was ≤14 points for MGCS, ≥7.5 points for NDS1, and ≥10 points for NDS2. At these cut-off values, MGCS had the highest PSe while NDS1 had the highest PSp (Table 1; Fig. S3). NDS1 appeared to be the most balanced prognostic test thanks to the highest simultaneous LR+ and LR- (25.8 and 0.12, respectively) (Table 1).
Table 1. Prognostic accuracy of the neurological assessment scales
Scale
|
AUROC (95% CI) [%]
|
Cut-off values
|
PSe (95% CI) [%]
|
PSp (95% CI) [%]
|
PPV (95% CI) [%]
|
NPV (95% CI) [%]
|
LR+ (95% CI)
|
LR- (95% CI)
|
MGCS
|
96.6
(92.7 – 99.9)
|
≤14
|
9/9
100 (70.1 – 100)
|
97/116
83.6 (75.8 – 89.3)
|
9 / 28
32.1 (17.9 – 50.7)
|
97 / 97
100 (96.2 – 100)
|
6.1
(4.0 – 9.2)
|
∞
|
NDS1
|
95.1
(86.9 – 99.9)
|
≥7.5
|
8/9
88.9 (56.5 – 98.0)
|
112/116
96.6 (91.5 – 98.7)
|
8 / 12
66.7 (39.1 – 86.2)
|
112 / 113
99.1 (95.2 – 99.8)
|
25.8
(9.6 – 69.4)
|
.12
(0.02 – 0.73)
|
NDS2
|
90.6
(79.6 – 99.9)
|
≥10
|
7/9
77.8 (45.3 – 93.7)
|
104/116
89.7 (82.8 – 94.0)
|
7 / 19
36.8 (19.1 – 59.0)
|
104 / 106
98.1 (93.4 – 99.5)
|
7.5
(4.0 – 14.3)
|
.25
(0.07 – 0.83)
|
Legend: MGCS – Modified Glasgow Coma Scale, NDS1 – Neuro Disability Scale designed by Smith et al. (2009), NDS2 – Neuro Disability Scale designed by Gonçalves et al. (2023), PSe – prognostic sensitivity, PSp – prognostic specificity, PPV – positive predictive value, NPV – negative predictive value, LR+ – likelihood ratio of the positive result, LR- – likelihood ratio of the negative result; ∞ – infinity, CI 95% – 95% confidence interval
Mid-term prognosis – probability of survival for 100 days
Excluding the 11 dogs that died during 1st week, 114 dogs were observed until day 100: 19/114 dogs died of MUO (16.7%) and 95 were alive on day 100 (83.3%). The S100 was 76.0% (95% CI: 67.8%–82.6%; 95/125 dogs) (Fig. 2).
None of the three neurological scales was significantly associated with the probability of surviving the period between 8 and 100 days. Two clinical variables, proved to be significantly and positively associated with a higher probability of survival in this period: unilateral brain lesions in MRI (p=0.004) and resolution of neurological signs (p<0.001) (Table S7). In the multivariable analysis, only spread of lesions in MRI was significantly associated with survival in this period – controlled for the treatment protocol used, dogs with unilateral lesions in MRI had approximately 12-fold higher odds of surviving than dogs with bilateral lesions in MRI (ORadj 12.5; 95% CI: 1.60–97.4; p=0.016) (Table S8).
Long-term prognosis – probability of survival for 1 year
Ninety five dogs were observed until day 365; 35/95 died of MUO (36.8%) and 60/95 were alive at the end of the 1st year (63.2%). S365 was 56.6% (95% CI: 47.1%–65.6%; 60/106 dogs) (Fig. S2). Survival rate calculated according to the most pessimistic and most optimistic scenario was 47.2% (95% CI: 38.8%–55.9%; 60/127 dogs) and 63.8% (95% CI: 55.1%–71.6%; 81/127 dogs), respectively, which virtually perfectly corresponded to the 95% confidence limits of S365.
In the univariable analysis, three factors proved to be significantly and positively associated with the period between 8 and 365 days after the onset of MUO signs: body weight of 3.2 kg and higher (p=0.002), presence of unilateral lesions in MRI (p=0.019), and resolution of neurological signs (p<0.001); while two other factors were significantly negatively associated with survival in this period: the presence of inflammatory pleocytosis in the CSF analysis (p=0.023) and impaired cerebellar function in NDS2 (p=0.045) (Table S9). The former factor could not be included in the multivariable analysis as CSF analysis was available only for 65/95 dogs. The latter factor was dichotomized into normal and impaired cerebellar function. In the multivariable analysis, dogs with body weight of 3.2 kg and higher on admission had approximately 6-fold higher odds of survival until day 365 (ORadj 6.49, 95% CI: 1.79–23.5; p=0.004), and dogs with unilateral lesions in MRI had approximately 3-fold higher odds of survival (ORadj 3.38, 95% CI: 1.03–11.1; p=0.045), while dogs with impaired cerebellar function had approximately 5-fold lower odds of survival (ORadj 0.21, 95% CI: 0.05–0.95; p=0.043) (Table S10).
Resolution of clinical signs
The resolution of clinical signs (both partial and complete) was strongly related to the probability of survival in the two aforementioned analyses (Table S7 and Table S9). However, on admission it is unknown if neurological signs will resolve so it is rather a clinical endpoint (next to survival) than a potential prognostic (risk) factor. Moreover, clinical improvement has a substantial influence on the patient’s quality of life, so the risk factors associated with the lack of neurological sign resolution were additionally separately investigated. The analysis included 108 dogs that survived for at least 1 month which was considered a time period sufficient for clinical improvement. Neurological signs did not resolve in 24/108 (22.2%) dogs and the remaining 84/108 (77.8%) dogs showed at least partial resolution of neurological signs (partial resolution in 40/84 dogs and complete resolution in 44/84 dogs). In the univariable analysis, the neurological scales themselves were not significantly associated with the resolution of neurological, but there were four elements of the neurological scales that were significantly associated with the lack of resolution of neurological signs: consciousness in MGCS, mentation in NDS1, and cerebral function and visual functions in NDS2. Moreover, the presence of unilateral lesions in MRI was significantly negatively associated with the lack of resolution of neurological signs (Table S11). The presence of unilateral lesions in MRI turned out to be the only factor significantly linked to a higher probability of resolution of neurological signs in the multivariable analysis (Table S12). Dogs with bilateral lesions on MRI had approximately 4-fold lower odds of resolution of neurological signs than did dogs with unilateral lesions (ORadj 0.28, 95% CI: 0.08–0.95; p=0.041).