Packed Red Blood Cell Transfusions in Dogs With Gastrointestinal Hemorrhage: 55 Cases (1999–2001)

Introduction

Retrospective analyses of packed red blood cell (PRBC) transfusions in veterinary medicine have found gastrointestinal (GI) hemorrhage to be a source of blood-loss anemia.^1–3 In one study, GI bleeding accounted for 6% of PRBC transfusions.³ That study may have underestimated the proportion of animals receiving transfusions for GI hemorrhage by classifying GI blood loss caused by thrombocytopenia or neoplasia separately. Multiple case reports have documented various causes of severe GI bleeding in dogs, including nonsteroidal anti-inflammatory drugs (NSAIDs),⁴ hypoadrenocorticism,⁵ factor XII and prekallikrein deficiency,⁶ and angiodysplasia;⁷ but a comprehensive review of the causes of GI hemorrhage in dogs has not been published.

The purpose of this study was to establish the frequency of severe GI hemorrhage requiring transfusion with PRBC, in the authors’ hospital, by broad etiological subcategories, including coagulopathy, intestinal pathology, and drug-related causes. Transfusion requirements, initial clinical laboratory data, hospitalization length, and patient outcome were also evaluated to attempt to identify prognostic indicators for duration of hospitalization, survival, and transfusion volumes based upon underlying disease.

Materials and Methods

The transfusion log at the Tufts University Foster Hospital for Small Animals was reviewed for cases of dogs with GI hemorrhage that received PRBC transfusions between January 1, 1999 and February 28, 2001. Any patients receiving whole blood or hemoglobin-based oxygen-carrying solutions for GI hemorrhage were excluded. The medical record was reviewed if the transfusion log entry reported that GI hemorrhage was the reason for transfusion, a disease process potentially associated with GI hemorrhage was cited, or a reason for transfusion was not listed. The medical record of each case was further reviewed to confirm GI hemorrhage as a factor in the decision to transfuse and to confirm that the anemia was consistent with blood loss (i.e., anemia and panhypoproteinemia). Dogs receiving transfusions for other reasons were excluded from the study.

The medical records of the eligible dogs were evaluated, and the following data was recorded: age, sex, breed, pre-transfusion packed cell volume (PCV) and total protein (TP), posttransfusion PCV and TP, number of separate transfusions, and total volume (mL) and units of PRBC administered during hospitalization. A unit was defined as containing 250 mL of PRBC. Additionally, history of NSAID or glucocorticoid use within 1 month prior to diagnosis, additional treatments during hospitalization, length of hospitalization, final diagnosis, and outcome were recorded. Gender distributions for all dogs admitted to the Foster Hospital for Small Animals were collected for comparison.

Laboratory tests submitted within 24 hours prior to transfusion were evaluated for red blood cell (RBC) morphology, platelet count, blood urea nitrogen (BUN), creatinine (Cr), albumin, and prolongations of the prothrombin time (PT) or activated partial thromboplastin time (APTT) >25% above the reference range. The BUN/Cr ratio, which when elevated has been suggested to be a marker of GI hemorrhage, was calculated for each dog and compared to a reference ratio of 20.⁸⁹ A commercially available fecal occult blood test^a was used when applicable to verify GI hemorrhage. This test can be influenced by diet, which the authors were not able to control given the retrospective nature of the study.

After review of the medical record, cases were divided into two etiological categories using the following criteria:

Results

For the 26-month study period in 1999 to 2001, 736 PRBC transfusions were administered to dogs, accounting for a total of 828.5 units. From these transfusions, 65 dogs were determined to have GI hemorrhage and were eligible for inclusion. Fifty-five of the 65 dogs were included; exclusions were due to incomplete medical records. These 55 dogs received a total of 84 separate transfusions of PRBC, accounting for 12% of all the PRBC transfusions administered in the authors’ hospital during the study period. These 55 dogs received a total of 81.8 units, which accounted for 10% of the PRBC units used during that same period. The criteria resulting in a decision to transfuse were not recorded in the medical record, although in the authors’ hospital, these criteria included clinical signs (e.g., tachycardia, weakness), change in PCV over time, and absolute PCV value.

Sixty-two percent of the dogs were male (25 neutered, nine intact), and 38% were female (19 spayed, two intact). The male to female ratio (1.6:1) was not significantly different from the overall hospital admissions ratio (1:1) during the same period (P=0.32). Forty-one purebreds (21 separate breeds) were represented, and there were 14 mixed-breed dogs. Breeds included the Labrador retriever (n=6), golden retriever (n=4), German shepherd dog (n=3), keeshond (n=3), Samoyed (n=3), American cocker spaniel (n=3), bichon frise (n=2), dachshund (n=2), border collie (n=2), boxer (n=2), and one each of greyhound, beagle, toy poodle, Irish setter, Boston terrier, English cocker spaniel, German shorthaired pointer, English springer spaniel, Cairn terrier, rottweiler, and Saint Bernard. The median age of the dogs was 8.7 years (range, 0.5 to 16.0 years), and median weight was 25 kg (range, 5.2 to 64.0 kg). The diagnosis of GI hemorrhage was supported by historical information in 42 dogs, physical examination findings in 46 dogs, endoscopy in four dogs, and fecal occult blood testing in one dog. Median duration of reported clinical signs of GI hemorrhage was 3 days (range, 1 to 365 days); two dogs had a more chronic history (1.5 and 12 months, respectively) of GI hemorrhage. Diagnostic imaging included abdominal ultrasound in 41 dogs and abdominal radiographs in 19. Five dogs had endoscopic evaluation and biopsy of the stomach and proximal duodenum. Seven dogs had exploratory celiotomies performed after identification of lesions with ultrasonography.

Complete blood cell counts (CBCs) were available for 54 of 55 dogs. In the two dogs with chronic GI bleeding, a microcytic, hypochromic anemia was noted; the remaining 52 had a normocytic, normochromic anemia. The mean hematocrit was 16.4±3.8% (reference range, 39% to 55%). The median platelet count was 130 × 10³ cells/μL (reference range, 200 to 550 × 10³/μL; range, 3 to 1,500 × 10³/μL); 36 (65%) dogs were thrombocytopenic, and four dogs had thrombocytosis including the two dogs with chronic GI bleeding. Twenty-nine dogs had PT and APTT performed, of which only two cases had prolongations >25% above the reference range.

Serum biochemical profiles were performed for 47 dogs prior to transfusion. The median BUN was 23 mg/dL (reference range, 8 to 29 mg/dL; range, 6 to 226 mg/dL), with 13 dogs having values greater than the reference range. The median creatinine concentration was 0.6 mg/dL (reference range, 0.6 to 2.0 mg/dL; range, 0.2 to 4.7 mg/dL), with four dogs having values higher than the reference range. The mean BUN/Cr ratio was 34 (range, 9 to 110). Thirty-seven (79%) dogs had a ratio >20. Eighty-seven percent of dogs were panhypoproteinemic, with a TP of <6.0 g/dL; the median TP was 4.2 g/dL (reference range, 6.0 to 7.8 g/dL; range, 3.3 to 6.6 g/dL).

Dogs with intestinal pathology comprised 69% (n=39) of the cases. Intestinal masses were found in seven dogs (13% of the 55 dogs), including leiomyosarcoma (n=4), adenocarcinoma (n=1), and lymphoma (n=1). The remaining case was not evaluated histopathologically, but on ultrasonographic examination, an eccentric, cavitated pyloric mass was described, and the dog was euthanized without further testing. Three dogs were identified with severe gastroenteritis. Gastrointestinal infiltrates were described histopathologically as severe neutrophilic gastritis with severe lymphoplasmacytic colitis; severe, diffuse, purulent enteritis with severe lymphoplasmacytic-eosinophilic colitis; and severe, erosive, eosinophilic and mononuclear enteritis. Another dog defecated multiple ingested foreign bodies and then had a large volume of severe GI hemorrhage. One young dog had severe GI hemorrhage attributed to infection with Ancylostoma and Toxocara spp. One dog with leiomyosarcoma had a concurrent history of aspirin administration.

Two dogs with intervertebral disk disease (IVDD), one cervical and one thoracolumbar, had severe GI hemorrhage. One was a dachshund that had received oral dexamethasone (0.35 mg/kg body weight, per os q 24 hours) chronically for 2 months. The other dog had received only one dose of dexamethasone (1.1 mg/kg body weight, intravenously) prior to presentation.

Two cases had been previously identified to have hepatic disease. One dog with severe cirrhosis had also received chronic glucocorticoids and NSAID (i.e., aspirin) therapy, while the other dog had a history of a surgically-ligated extrahepatic shunt and received two doses of an NSAID (i.e., carprofen^c) prior to presentation. Three dogs with chronic renal failure were transfused due to GI bleeding. Two of the three cases had a concurrent history of NSAID (i.e., etodolac^d and aspirin) use prior to presentation. Finally, a dog with hyperadrenocorticism that was receiving aspirin also developed GI bleeding severe enough to warrant transfusion.

In 14 dogs, the use of anti-inflammatory drugs was suspected as the sole cause of GI hemorrhage. Dogs received aspirin (n=5), naproxen (n=3), carprofen (n=2), carprofen and aspirin (n=1), piroxicam and aspirin (n=1), etodolac (n=1), prednisone (n=1), and dexamethasone (n=1). The three dogs that received naproxen were the only accidental ingestions and the only cases in which the anti-inflammatory dosages exceeded published clinical guidelines.¹⁰ Dosage schedules varied in length from 1 day to 5 years prior to the onset of clinical signs, while some dogs were only intermittently administered medication. Two dogs received chronic treatment (>2 months) with both aspirin and prednisone.

Multifactorial diseases were represented in the remaining five cases of dogs with intestinal pathology. One dog had evidence of pancreatitis, gastroenteritis, thrombocytopenia, splenic infarction, and abdominal effusion. Another dog on chemotherapy (i.e., glucocorticoids and azathioprine) for systemic mastocytosis with thrombocytopenia was also transfused for GI hemorrhage. An abdominal mass with neoplastic effusion was found in another dog with pancreatitis and prolongations in the PT and APTT, but the dog was euthanized prior to complete diagnosis. A round cell neoplasia infiltrating the liver, spleen, and bone marrow was documented in another case with thrombocytopenia. A dog previously diagnosed with IMT and treated with prednisone and azathioprine presented with a platelet count within the reference range (247 × 10³/μL) and also had suspected pancreatitis, toxic hepatopathy, and GI hemorrhage.

Sixteen (31%) dogs were identified as having primary coagulopathies; all of these were diagnosed as having IMT (median platelet count, 5.5 × 10³/μL; range, 2.0 to 11 × 10³/μL). Dogs with IMT were significantly younger (median, 8.0 years; range, 3 to 10 years) than nonIMT dogs (median, 11.0 years; range, 0.5 to 16 years; P<0.01). Fifty percent of the dogs with IMT were female, and four (25%) were cocker spaniels. One dog was treated for IMT with dexamethasone, prednisone, and a whole-blood transfusion by the referring veterinarian 4 days prior to presentation. Another dog with IMT received one dose of aspirin prior to presentation.

A median of 11.3 mL/kg body weight (range, 5.7 to 27.7 mL) of PRBCs was administered to these 55 dogs for their first transfusion due to GI hemorrhage. The majority (89%) of dogs were transfused within 24 hours of initial hospital presentation. A median total volume of 13.9 mL/kg body weight (range, 5.6 to 78.1 mL) was administered per dog during the entire hospitalization period. Dogs with IMT received a significantly (P<0.01) larger total volume of PRBCs than nonIMT (i.e., intestinal pathology) dogs [Figure 1]. Dogs with IMT were also more likely to receive multiple transfusions than nonIMT dogs (P<0.03); the median number of transfusions for IMT dogs was two, while nonIMT dogs received a median of one (both ranged from one to four transfusions).

Gastroprotectants were administered in 51 (93%) dogs, including H₂ blockers, proton pump inhibitors, prostaglandin analogues, and coating agents. Forty-one (75%) dogs received two or more gastroprotectants. Synthetic colloid fluid therapy^e was administered in 13 (23%) dogs. Fresh-frozen plasma (FFP) was administered to nine (16%) dogs during their hospitalization. Recorded rationale for FFP administration in these dogs included oncotic support (n=7) and coagulation factors (n=1); no reason was cited for the remaining dog. No dogs received a hemoglobin-based oxygen-carrying solution during the study period for the treatment of GI hemorrhage per the blood bank log; in the authors’ hospital, PRBC transfusions are the oxygen-carrying fluid of choice.

Dogs were hospitalized for a median of 4 days (range, 1 to 23 days). Sixteen (29%) of the 55 dogs in the study did not survive to discharge; 12 were euthanized and four died. Significant differences in hospitalization or outcome were not found between groups.

Discussion

Gastrointestinal hemorrhage and subsequent development of moderate to severe anemia accounted for 12% of the PRBC transfusions administered in a 26-month study period at Tufts University. This number is greater than that previously reported in the veterinary transfusion literature,³ but includes all causes of GI bleeding, including thrombocytopenia and intestinal neoplasia. Therefore, GI hemorrhage should be considered as an important cause of blood-loss anemia.

Severe thrombocytopenia was found to be a potentially important factor in the development of GI hemorrhage in this study. Dogs with IMT represented over one-third of dogs receiving PRBC for GI hemorrhage. The exact mechanism for GI hemorrhage in dogs with severe thrombocytopenia is unclear. One theory is that without thrombocytes to repair vascular defects, vessels in the GI mucosa may bleed and, given the large surface area and vascularity of the intestinal tract, large blood losses could occur.¹¹

The incidence of GI hemorrhage in dogs with IMT is not currently known, but it is thought to be common.¹² Given the findings of the current study, dogs with IMT exhibiting GI hemorrhage appear to commonly receive multiple PRBC transfusions. One theory would be that GI hemorrhage secondary to thrombocytopenia will not resolve without restoration of a normal platelet count. With current therapeutics and without platelet transfusions, this will likely be after a minimum of 4 to 5 days.¹² Dogs with ulcerative GI diseases may recover faster. It should also be recognized that some dogs with IMT likely may have had substantial losses at other sites in the body, such as the subcutaneous spaces, urinary bladder, or other highly vascular epithelial sites.

A number of conditions that potentially cause GI ulcers were seen in this study, including hepatic disease, gastroenteritis, renal disease, pancreatitis, systemic mastocytosis, IVDD, and steroidal and nonsteroidal anti-inflammatory drug administration. A previous retrospective study of GI ulcerative disease in dogs found that hepatic disease and NSAID therapy were the two most common underlying risk factors for ulcerative disease.¹³ Similarly, in a review of dogs and cats with spontaneous gastroduodenal perforation, commonly identified concurrent diseases or therapeutics included hepatic disease, IBD, gastroduodenal and nongastroduodenal neoplasia, and NSAID administration.¹⁴ Several reports document severe GI hemorrhage in dogs following dosages of NSAIDs in the published therapeutic range.⁴¹⁵ The timing of presentation of these dogs for clinical GI hemorrhage can vary in length, from after a few doses to a year of administration.⁴¹⁵ These variable temporal relationships also were noted in this study population.

The risk of NSAID therapy in dogs and humans also appears to be linked to comorbid illnesses,¹³¹⁶ as was also noted in this population. Comorbid illnesses detected in this study included IVDD as well as renal and hepatic diseases. It would appear that NSAID therapy in dogs with preexisting diseases with GI ulcerogenic potential should be approached cautiously.

Corticosteroid administration is an established risk factor for GI ulcer formation in humans,¹¹ but the relationship has been more difficult to document in dogs. Studies have documented GI ulcerative lesions in dogs treated with various corticosteroids; however, these dogs had significant preexisting diseases with ulcerogenic potential, such as IVDD, hepatic disease, and systemic mastocytosis.^1317–20 Corticosteroids may be important in the development of GI hemorrhage in dogs with comorbid illness or in those concurrently receiving NSAIDs.¹³²¹

As seen in an earlier study,⁹ the BUN/Cr ratio was not specific or sensitive in identifying severe GI hemorrhage. Although the BUN/Cr ratio was moderately elevated for most dogs, a significant proportion (21%) of dogs with severe GI hemorrhage had a normal or low ratio. This may be due to the fact that this ratio can be altered by many factors.

The overall mortality rate of 29% seen in the current study was less than that seen in previous studies for dogs receiving PRBC transfusions in general (53%¹ and 39%³) and, more specifically, for severe GI hemorrhage (45%³). In the human literature, an overall mortality rate of 7% to 10%²² has been reported for patients with severe GI hemorrhage.

Limitations of this study include those inherent in a retrospective study, such as completeness of medical record information and availability. Some previously documented causes of severe GI hemorrhage were not seen during the study period, such as anticoagulant rodenticide toxicity and hypoadrenocorticism, but animals with these problems have been treated with PRBC at the authors’ institution outside the studied period. Additionally, many studies in humans have documented the prevalence of low-grade GI hemorrhage in a variety of critically ill patients;²³ but, due to the retrospective nature of this study, the authors were not able to retrieve data to assess GI hemorrhage in dogs that did not receive a PRBC transfusion.

Conclusion

Anemia resulting from GI hemorrhage accounted for 12% of the PRBC transfusions at the authors’ institution during the 26-month study period. Immune-mediated thrombocytopenia was a relatively common cause of GI hemorrhage, and dogs with IMT received significantly more blood than dogs with intestinal pathology leading to GI hemorrhage. Anti-inflammatory drug use played an important role in GI hemorrhage, especially when coupled with a GI ulcerogenic comorbid illness.