Included cases
A total of 135 cases were included in the study and further divided into two groups: i) 115 cases were classified as idiopathic peritonitis cases without an identifiable cause of the peritonitis apparent during the duration of hospitalization, ii) 20 cases had confirmed NSII of the pelvic flexure, determined by abdominal surgery and pathological assessment of the resected intestine or full body necropsy.
Excluded cases
Fifteen cases were excluded from the main analyses, as they did not meet the set inclusion criteria for case selection. Of these, two cases had suspected parasitic lesions involving the jejunum. One case was suspected of NSII at the pelvic flexure, but the resected intestine was not sent for pathological assessment, and one case had a small infarction in the colon that was not resected. The remaining 11 cases had rectal examinations and/or rectal ultrasonographic findings suggestive of a mass, thickened intestinal wall at the pelvic flexure, adhesions and/or suspected thrombus formation in the mesenteric vessels, without confirmative diagnosis. Although unconfirmed, it was acknowledged that these latter horses could represent NSII cases that responded to medical treatment and the progression of clinical signs and outcome in this group was therefore recorded. In addition, all 15 horses were included as a separate group (unconfirmed NSII cases) in the cluster analysis.
Exploratory and descriptive analyses using cluster analyses
In the cluster analyses (Figure 1) clinical diagnosis clustered closely to the parameters leucocyte count, neutrophil count, specific rectal findings (rectal mass/adhesions or thickened intestinal wall) as well as the outcome parameters euthanized and days at clinic (see box “a”, Figure 1). The pattern suggested that confirmed NSII was associated with lower leucocyte and neutrophil counts. The majority of individuals with confirmed NSII showed high similarity and clustered together (see box “b”, Figure 1.). The common aspects of the individuals in cluster “b” were that they presented primarily during the winter months with a rectal mass/adhesions or thickened intestinal wall and a poor outcome. Cluster “b” also included a subcluster (see box “c”, Figure 1) with unconfirmed NSII cases. In this group, about half of the horses survived to discharge and it appeared to include slightly older horses. The confirmed cases (cluster “b”) also appeared to divide into subclusters based on clinical signs. Confirmed cases with more severe colic at home and at admission, with signs of endotoxaemia, and that were often taken to surgery clustered together (see box “d”, Figure 1). Most of the remaining horses with unconfirmed or confirmed NSII were found in a larger cluster including horses diagnosed with idiopathic peritonitis (see box “e”, Figure 1). The horses with unconfirmed NSII differed from the horses in cluster “b” in that they presented earlier in the year (during autumn), survived to discharge and had less elevated total protein in the peritoneal fluid as well as less decreased leukocyte count in the blood (neutrophil counts were missing for the majority of these horses).
Demographic data
In the idiopathic peritonitis group, Warm-blooded horses (including Thoroughbreds/Arabians) were most common (54 %), followed by pony breeds (17 %), Icelandic horses (17 %), Standardbred trotters (6 %), and Cold bloods (5 %). One horse was of unknown breed. Similarly, Warmbloods predominated in the NSII group (72 %), with 17 % Cold bloods, one Icelandic horse, one of pony breed and one Arabian. In two cases in the NSII group the breed was not recorded. The idiopathic case group consisted of 50 % geldings, 42 % mares and 8 % stallions. Of the NSII confirmed cases, 65 % were geldings, and 30 % mares, with one stallion. The mean (± SD) and median age in the two groups were as follows: idiopathic 12.6 (± 6.1) and 12 years; NSII 11.7 (± 4.8) and 10.5 years. There were no significant differences in demographic data between groups (Suppl. Table 1a).
Previous medical history and admission data
No statistical differences in relevant medical history data was found (Suppl. Table 1b). Within both groups, the number of cases that had received anthelmintic treatment within six months of presentation was similar to the number not treated within this time-frame, although this data was missing in a large number of cases in the idiopathic case group (37 %). In total, seven cases with confirmed NSII lesions had been treated with an anthelmintic drug within six months of admission. Variables with a significant difference between the two study groups are presented in Table 1. There was a significant effect of season (p = 0.001), with horses diagnosed with NSII more often presenting during the autumn/winter months (Table 1), which was, as shown in the cluster analysis (Figure 1), due to NSII cases presenting primarily during the winter months (Dec-Feb). The predicted probability of a horse being diagnosed with NSII increased markedly over the winter months, compared to the summer season (Figure 2). Horses later confirmed with NSII more often presented with obvious colic signs requiring analgesics (p = 0.005), but without fever (p = 0.027) (Table 1). However, this was not clear from the cluster analysis, due to the few number of cases with grade 3 colic. By case definition, no palpable masses, thickened intestine or adhesions were recorded in the idiopathic cases. However, such rectal findings were present in 65 % of horses confirmed with NSII lesions and showed close association with both confirmed and unconfirmed NSII cases in the cluster analysis. A normal rectal examination was significantly more common in the idiopathic case group (p < 0.001), with only one case diagnosed with NSII having no abnormal rectal findings. Trans-abdominal ultrasound was performed in 18 horses in the idiopathic peritonitis group, showing an increase in the amount of peritoneal fluid in all horses and in one case, also a mildly thickened small intestinal wall. No rectal ultrasound examination was performed in this group. In the NSII group, four horses had trans-abdominal ultrasound examinations, seven horses had rectal ultrasound examinations, and one horse had both types of examinations. On the trans-abdominal ultrasound, findings included increased amount of peritoneal fluid (n = 2), thickened small intestine (n = 3) and thickened colon wall (n = 1). All rectal ultrasound examinations showed a thickened colon wall, a localized mass involving or adjacent to a thickened colonic wall and/or suspected adhesion formation. Rectal ultrasounds were only performed if there had been abnormal rectal findings suggestive of a mass, thickened colon wall or adhesion formation. Remaining clinical data obtained at admission did not differ between groups (Suppl. Table 1c).
Table 1. Variables with significant differences between study groups.
Variable
|
Idiopathic, n (%)
|
NSII, n (%)
|
P-value
|
OR
|
CI (95%)
|
Season at presentationa
|
Spring/Summer
|
44 (38)
|
2 (10)
|
Reference variable
|
Autumn/Winter
|
71 (62)
|
18 (90)
|
0.019
|
5.58
|
1.51-36.16
|
Colic at admissiona
|
|
No colic at admission
|
35 (30)
|
4 (20)
|
Reference variable
|
Colic grade at admission
|
|
Grade 1
|
70 (61)
|
12 (60)
|
0.509
|
1.50
|
0.48-5.67
|
Grade 2
|
8 (7)
|
0 (0)
|
0.991
|
N/A
|
N/A
|
Grade 3
|
2 (2)
|
4 (20)
|
0.005
|
17.50
|
2.62-161.78
|
Fever at admission (≥ 38.5)a
|
54 (93)
|
4 (20)
|
0.027
|
0.27
|
0.08-0.80
|
Missing
|
2
|
0
|
|
|
|
Rectal examinationa
|
|
|
|
Mass/adhesions
|
0 (0)
|
13 (65)
|
< 0.001
|
Part of case definition
|
Normal rectal exam
|
54 (51)
|
1 (5)
|
< 0.001
|
0.05
|
0.00-0.35
|
Missing
|
8
|
0
|
|
|
|
WBC (median [IQR])b
|
7.08 [5.50, 9.62]x109/L
|
3.50 [3.05, 5.01]x109/L
|
< 0.001
|
0.04†
|
0.28-0.64
|
Missing
|
7
|
5
|
|
|
|
Neutrophils (median [IQR])b
|
4.95 [3.60, 7.22]x109/L
|
2.40 [1.40, 3.19]x109/L
|
< 0.001
|
0.55†
|
0.38-0.75
|
Missing
|
7
|
4
|
|
|
|
Elevated fibrinogen levela
|
18 (20)
|
8 (53)
|
0.012
|
4.70
|
1.28-17.19
|
Missing
|
23
|
5
|
|
|
|
Peritoneal total protein level (median [IQR])b
|
40.00 [34.00, 50.00] g/L
|
50.00 [40.00, 60.00] g/L
|
0.011
|
1.38‡
|
1.10-1.78
|
Missing
|
18
|
3
|
|
|
|
a Fischer’s exact test, b Kruskal-Wallis rank sum test, † for every increase of 1.0 x 109/L, ‡ for every increase in 5.0 g/L
Laboratory parameters and parasitic diagnostics
All blood samples included were obtained prior to surgical intervention. Haematocrit was excluded from the analysis since different haematological analysers were used rendering non-comparable values between cases. No horse in either study group presented with a haematocrit above 50 %. Elevated fibrinogen levels on first sampling occasion were significantly more common in the NSII group (Table 1, p = 0.012), but cluster analyses showed idiopathic cases with similar results (Figure 1). Leukopenia, characterized by a neutropenia, was significantly more frequent in the NSII group compared to horses diagnosed with idiopathic peritonitis (p < 0.001) (Suppl. Figure 1a, 1 b) and cluster analyses showed close association with final diagnosis. The mean protein level in the peritoneal fluid was higher in horses diagnosed with NSII as compared to the idiopathic case group (p = 0.011). However, as seen in the cluster analysis, there were horses in the idiopathic group with comparable levels. No other laboratory parameters were different between the two study groups (Suppl. Table 1d).
Antimicrobial treatment regimes
All but one case in the idiopathic peritonitis group received antibiotic treatment. The horse that was not treated with antibiotics had peritoneal fluid with only a mildly elevated nucleated cell count (12,250 x 106) and was discharged within five days from admission. The remaining 114 horses were treated with either intravenous penicillin G sodium (51 %), penicillin G sodium and gentamicin (39 %), penicillin G sodium and trimethoprim sulphonamide (5 %) or various other combinations (4 %). There was a difference in initial treatment regimen between the three hospitals, with the majority of horses at hospitals one and three treated with penicillin G sodium only (76 % and 69 %, respectively), whereas the broad-spectrum antibiotic combination of penicillin G sodium and gentamicin dominated at hospital two (78 %). In the NSII group, four of cases were treated with penicillin G sodium only, all at hospital one. Six horses were initially treated with intravenous penicillin G sodium only, with the addition of either gentamicin (n = 5) or trimethoprim sulphonamide (n =1) due to continuing fever (n = 4) or decision for surgery (n = 2). The remaining horses were treated with a combination of penicillin G sodium and gentamicin from admission.
Progression of colic signs and fever during hospitalisation
In the idiopathic case group, there was a successive decline in the percentage of cases with fever (≥ 38.5°C) during the first 48 hours of antimicrobial treatment. In contrast, approximately one third of horses subsequently diagnosed with NSII were still febrile 48 hours after treatment initiation. However, in the NSII group, seven cases had surgical intervention and three cases were euthanized within this time-frame. Obvious signs of colic (grade 2 or 3) were uncommon in the idiopathic case group, with a further reduction in the number of cases showing colic over the first 48 hours of medical treatment. In contrast, the proportion of cases with colic in the NSII group increased over the first 48 hours of medical treatment, together with a progressive reduction in the number of horses on medical treatment due to surgery or euthanasia. For a summary of colic signs and fever during the first 48 hours of medical treatment, see Figure 3.
The majority of horses presenting with peritonitis and a palpable mass, adhesions or suspected thrombosis (confirmed and unconfirmed NSII cases) that were febrile or had obvious signs of colic 48 hours after initiated antimicrobial treatment, were either euthanized or taken to surgery (Figure 4). The remaining eight cases were treated medically and survived to discharge.
Surgical findings
One horse in the idiopathic case group was subjected to surgery due to a right dorsal displacement of the ascending colon without vascular compromise. In the confirmed NSII group, all twelve cases where surgery was performed had marked changes involving a segmental portion of the pelvic flexure or the juncture of the dorsal colon and the pelvic flexure, with a clear demarcation between healthy and unhealthy bowel without signs of strangulation (Figure 7, panel A). The compromised area of colon measured from five to 40 cm in length. In eight cases, thickening of the mesentery and/or enlargement of mesenteric lymph nodes were described. Adhesion formation between the pelvic flexure and the cecum and abdominal wall was found in three cases, of which one had a focal perforating lesion. Thrombosis formation in a ventral mesenteric vessel was observed in one horse.
Outcome
All cases in the idiopathic peritonitis group were discharged after varying lengths of medical treatment, or medical treatment and surgical correction of a colon displacement in one case, with a survival to discharge rate of 100%. Most horses were discharged within 6-15 days of admission (72 %). Thirty-two horses (28 %) were treated with antibiotics after discharge at home for a duration of two to eight days, although in 50 % of cases treatment length at home was unknown.
In the NSII group, twelve horses (60 %) had surgery performed between one to seven days after admission, with resection of the affected colon segment in ten cases. One horse was euthanized at surgery due to a ruptured bowel and another due to owner declination of colon resection. Five of the ten cases where colon resection had been performed survived to discharge, giving a short-time survival rate of 50 % after colon resection. These cases had surgery performed at admission (n = 2), three (n=1), four (n =1) and seven days (n =1) after presentation. One case that survived to discharge was subsequently euthanized nine months after surgery, due to stricture formation at the anastomosis site. Another five cases were euthanized between 4 to 25 days after colon resection due to persistent colic signs or persisting septic peritonitis. Two horses in addition developed wound infections, which, however, were not the reason for euthanasia. Eight cases, where surgery was declined by the owner based on the guarded prognosis, were euthanized one to four days from admission. A summary of the outcome in the NSII group is shown in Figure 5. Horses with rectal findings suggestive of NSII had a survival rate to discharge of 38 %, as displayed in Figure 6.
Pathology
Necropsy
In the NSII group, 9/20 (45 %) of the horses had the diagnosis confirmed at necropsy, with the remainder confirmed after pathological examination of the surgically resected portion of intestine. In all cases, as defined by the inclusion criteria, the intestinal lesions were localized in or near the pelvic flexure. In most necropsied cases (7/9), a well-defined segmental area of transmural necrosis with varying degrees of eosinophilic inflammation of the pelvic flexure was demonstrated, in one case resulting in a perforating intestinal lesion. In one horse, the necropsy was performed after resection, where the resected intestine had not been sent for histopathological assessment. However, necrosis and ulcerations were observed at the anastomosis site and the caudal and cranial mesenteric roots showed severe lesions with thrombosis and multiple eosinophilic granulomas compatible with injuries secondary to S vulgaris larval migration. The segmental lesion in the final horse was described as a 40 cm long segmental haemorrhagic infarction with multiple ulcerations. The length of the affected portion of colon, when noted, was between 5 and 40 cm. Endarteritis of the cranial or caudal mesenteric roots and/or major mesenteric branches was found in all of horses, with thrombus formation recorded in seven cases. In six cases, S. vulgaris larvae were found within the arterial and/or intestinal lesions (Figure 7, panel C). Reactive hyperplasia and haemorrhage of the mesenteric lymph nodes or eosinophilic lymphadenitis was observed in four of necropsied cases. In four cases, fibrinous to fibrous adhesions were found. In two of these horses, there was severe adhesion formation between the colon ascendens, the abdominal wall and colon descendens, involving the uterus in one case (Figure 7, panel B). An acute, subacute, purulent or fibrinopurulent peritonitis was recorded in all but one case, in which the necropsy was performed five days after resection of the affected portion of intestine.
One horse that underwent colon resection, with histopathological examination of the resected tissue, was subsequently sent for necropsy nine months later, after euthanasia, due to repeated colon impactions. A stricture formation was demonstrated at the anastomosis site, but there were no macroscopic lesions in the cranial or caudal mesenteric arteries or its branches. Histological evaluation of the mesenteric arteries was not performed.
Resected intestine
The remaining eleven cases in the NSII group, all underwent exploratory laparotomy with findings as described in the previous section. Colon resection was performed in all cases (one after euthanasia to confirm diagnosis) and histopathological evaluation of the resected intestine showed severe transmural lesions, with variable inflammatory, fibrinosuppurative to necrotic changes, in most cases with substantial eosinophilic inflammation. In nine cases, vascular lesions with thrombus formation were recorded. One of these cases was identified during surgery with thrombosis in a ventral mesenteric artery and in four cases, parasites consistent with S. vulgaris larvae were found.