Criteria for Selection of Cases

Medical records at the Purdue University Veterinary Teaching Hospital were searched to identify dogs with histologically confirmed lymphoma that were treated with chemotherapy from 1993 through 2006. Dogs were included in the study if clinical staging of the neoplasia had been performed and hematologic and follow-up data were available.

Procedures

Data collected from the medical records included breed; age; weight; sex and neuter status; anatomic form of the lymphoma; and initial hematologic data, including Hct, RBC count, mean corpuscular volume, MCHC, and mean hemoglobin concentration. Other clinicopathologic data, including mean serum albumin concentration and alanine aminotransferase activity, were also noted. Additional information collected from the medical records included the date on which chemotherapy was started, chemotherapy protocol used, clinical response, date of last follow-up, and date of death. On the basis of physical examination findings and results from diagnostic imaging and cytologic examination of bone marrow specimens, a clinical stage and substage were assigned to each dog by use of the WHO criteria for canine lymphoma.¹⁹

Dogs with Hct ≤ 37% were considered anemic; 37% represented the lower reference limit for Hct used by the hospital's clinical pathology laboratory. On the basis of Hct values, anemia was further characterized as mild (Hct, 30% to 37%), moderate (20% to 29%), or severe (13% to 19%).²⁰ Clinical response to treatment was categorized in terms of complete response (no evidence of disease), partial response (≥ 50% reduction in tumor volume), stable disease (< 50% reduction in tumor volume), and progressive disease (≥ 50% increase in tumor volume or development of new lesions). During data analysis, partial response, stable disease, and progressive disease were considered together as a variable designated as not complete response. Unfortunately, reticulocyte counts were not available for most of the dogs included in the study.

Survival time was defined as the interval between the date on which chemotherapy was started and the date of death. Observations were right censored if a dog died from causes other than lymphoma or was lost to follow-up.

Statistical analysis—Categoric data were compared between groups by use of a χ² test, unless > 20% of the expected values in the contingency table were < 5, in which case a Fisher exact test was used. Continuous variables were compared between groups by use of a Student t test. Logistic regression was used to model the effect of predictor variables on clinical response,²¹ and Cox-proportional hazard regression was used to model the effect of predictor variables on survival time.²² Estimated survival curves were generated by use of the product-limit method²³; survival curves were compared between groups by use of the log-rank test.²⁴ All statistical analyses were done with computer software.^a Values of P < 0.05 were used to indicate significance. Data are expressed as mean ± SD unless otherwise stated.

Results

Initially, 110 medical records from dogs with lymphoma were identified, but only records from 96 dogs contained sufficient information regarding anemia, treatment, and follow-up for inclusion in the study. Of these 96 dogs, 40 (41.7%) were anemic and 56 (58.3%) were not anemic at the time of diagnosis. As part of the clinical staging, cytologic examination of a bone marrow aspirate was performed for 92 (96%) dogs. In each of these dogs, bone marrow specimens were collected from a single site, including the ileum (n = 25; 28%) and humerus (53; 57%); the site of collection was unknown for 14 (15%) dogs. Of the 4 dogs from which bone marrow samples were not collected, all did not have anemia. Regardless of the severity of anemia, all dogs were classified as having a nonregenerative (normocyticnormochromic) anemia on the basis of red cell indices. Reticulocyte data were not available for most dogs; therefore, those data were not considered in the classification of anemia or used in statistical analysis.

The dogs with and without anemia were similar with respect to age, weight, distribution of breeds, sex and neuter status, anatomic form of the lymphoma, and evidence of bone marrow involvement (Table 1). Concurrent disease was not an important cause for anemia in the 40 affected dogs. Among the anemic dogs, only 2 (5%) had high serum alanine aminotransferase activity (333 U/L and 272 U/L; reference range, 3 to 69 U/L); however, both dogs had ultrasonographic evidence of lymphoma-induced changes in the liver. None of the anemic dogs were azotemic at the time of diagnosis. All abnormalities detected via thoracic and abdominal radiography and abdominal ultrasonography were consistent with lymphoma; diagnostic imaging did not reveal evidence of other disease processes. Pituitary gland–dependent hyperadrenocorticism was diagnosed in 1 (2.5%) anemic dog before the diagnosis of lymphoma. Incomplete medical record data precluded accurate substage classification of 5 dogs.

Significantly (P = 0.005) more dogs that were not anemic at the time of diagnosis were characterized as having a complete response to treatment, compared with findings in the dogs that were anemic. The distributions of WHO stage and substage, serum albumin concentration, and measures of anemia differed significantly between the anemic and nonanemic dogs (Table 1). On the basis of cytologic evidence of lymphoma infiltration into the bone marrow, 27 of the dogs evaluated were classified as WHO stage V; 14 (52%) of those dogs were anemic at the initial evaluation. Eleven dogs were also classified as WHO stage V on the basis of the presence of extranodal lymphoma; 5 (45%) of those dogs were anemic at the initial evaluation. Among the 19 dogs with stage V lymphoma that were anemic at the initial evaluation, mean Hct of dogs with bone marrow involvement was not significantly (P = 0.267) different than the mean Hct of dogs without evidence of bone marrow involvement (29.8 ± 3.53% and 25.8 ± 6.42%, respectively).

Among the 33 dogs that received treatment involving doxorubicin, 8 (24%) were treated with a standard combination chemotherapy protocol,²⁵ 22 (67%) were treated with a short induction-only protocol,²⁶ 2 (6%) were treated with doxorubicin and L-asparaginase, and 1 (3%) was treated with doxorubicin and an anti-lymphoma monoclonal antibody.²⁷ Of the 63 dogs treated without doxorubicin, 31 (55%) were treated with the COP protocol,²⁸ 22 (39%) were treated with the COP protocol and L-asparaginase,²⁹ 7 (5%) were treated with prednisone alone, and 3 were treated with the COP protocol and actinomycin-D. Significantly (P = 0.013) more dogs without anemia were treated with a chemotherapy protocol that included doxorubicin, compared with anemic dogs.

Table 1—

Characteristics of 96 dogs with lymphoma that were anemic (Hct le; 37%) or were not anemic (Hct > 37%) at the time of diagnosis. Values are given as mean ± SD or number of dogs (%).

View Table

Table 2—

Univariate analysis of predictors of complete response in 96 dogs with lymphoma that were treated with chemotherapy.

View Table

Table 3—

Univariate analysis of predictors of overall survival time in 96 dogs with lymphoma that were treated with chemotherapy.

View Table

Among the 40 anemic dogs, 36 (90%) remained anemic following chemotherapy; only 4 (10%) dogs had Hct values within reference range after 6 to 10 weeks of chemotherapy. Of those 4 dogs, 2 had lymphoma classified as WHO stage V with extranodal involvement but no evidence of bone marrow infiltration, and 2 had multicentric lymphoma classified as WHO stage IV. Because of the small number of dogs, it was not possible to determine the effect of the return of Hct to within reference limits after chemotherapy on survival time.

Univariate analysis revealed that RBC count, hemoglobin concentration, Hct, anemia, and lack of doxorubicin in the chemotherapy protocol were significantly associated with a decreased likelihood of a complete remission (Table 2). Dogs without anemia were 4 times as likely as dogs with anemia to have a complete response to chemotherapy. Each 1% increase in Hct was associated with a 7.6% increase in the odds of a dog achieving complete remission with treatment. Dogs that were not given doxorubicin in their chemotherapy protocol were considerably (75%) less likely to achieve complete remission than dogs treated with a protocol including doxorubicin. Because of the common use of Hct as an indicator of anemia and the strong correlation of Hct with hemoglobin concentration and RBC count, Hct was the only anemia-related variable used in multivariate model construction. Both Hct and chemotherapy protocol remained significant (P = 0.011 and P = 0.029, respectively) predictors of a complete response when included in the multiple logistic regression model.

Univariate analysis of the effects of continuous variables on overall survival time revealed that only RBC count, hemoglobin concentration, Hct, and MCHC were significantly associated with survival (Table 3). Univariate analysis of categoric variables revealed that anemic dogs had a significantly (P = 0.008) shorter median survival time (139 days), compared with survival time of nonanemic dogs (315 days; Figure 1). There was no significant (P = 0.943) difference in the median survival time between dogs that had doxorubicin included in their chemotherapy protocol (271 days) and those that did receive doxorubicin (297 days; Table 4). Because of the strong correlation between RBC count, hemoglobin concentration, Hct, and MCHC, multivariate analysis was not done.

View Full Size

Estimated survival curves for 96 dogs with lymphoma that were treated with chemotherapy. At the time of diagnosis, 40 dogs were anemic (Hct ≤ 37%; broken line) and 56 dogs were not anemic (Hct > 37%; solid line). The vertical marks are right-censor marks. The curves are significantly (P = 0.008) different.

Citation: Journal of the American Veterinary Medical Association 231, 12; 10.2460/javma.231.12.1836

• Download Figure
• Download figure as PowerPoint slide

Median survival times among dogs grouped by severity of anemia were calculated. Median survival times for dogs with mild (n = 14), moderate (20), or severe (6) anemia was 171, 139, and 34 days; median survival time for the 56 that were not anemic was 315 days. The median survival times of dogs with mild and severe anemia were significantly (P = 0.047 and P < 0.001, respectively) shorter than the median survival time for nonanemic dogs.

Table 4—

Life table survival analysis for 96 dogs with lymphoma that were treated with chemotherapy.

View Table

Table 5—

Characteristics of 48 dogs with multicentric lymphoma that were (Hct le; 37%) or were not (Hct > 37%) anemic at the time of diagnosis matched on the basis of clinical stage and chemotherapy protocol in a case-control subgroup analysis. Values are given as mean ± SD or number of dogs.

View Table

Because of concerns about the differences in chemotherapy protocols used for dogs with and without anemia, data from 24 anemic dogs with multicentric lymphoma were compared with data from 24 nonanemic dogs with multicentric lymphoma. These dogs were matched on the basis of WHO clinical stage and chemotherapy protocol. Within this case-control subset, the anemic and nonanemic dogs were similar with respect to age, weight, sex, distribution of breeds, and clinical substage of neoplasia (Table 5). Univariate analysis revealed that the anemic dogs had a significantly (P = 0.002) shorter median survival time (101 days), compared with survival time of nonanemic dogs (284 days; Figure 2). In this group of dogs, Hct at the time of diagnosis was significantly (P = 0.038) associated with the likelihood of a complete remission (odds ratio, 1.066; 95% confidence interval, 1.004 to 1.132).

View Full Size

Estimated survival curves for 48 dogs with multicentric lymphoma that were treated with chemotherapy. At the time of diagnosis, 24 dogs were anemic (Hct ≤ 37%; broken line) and 24 dogs were not anemic (Hct > 37%; solid line). These dogs were matched with respect to WHO stage of neoplasia and chemotherapy protocol. The vertical marks are right-censor marks. The curves are significantly (P = 0.002) different.

Citation: Journal of the American Veterinary Medical Association 231, 12; 10.2460/javma.231.12.1836

• Download Figure
• Download figure as PowerPoint slide

Discussion

Anemia in veterinary cancer patients has been considered of limited clinical consequence.⁴ However, a survey of more than 15,000 humans with neoplastic disease revealed that fatigue related to anemia had a more severe negative impact on their quality of life than any treatmentor cancer-related complication.¹ Approximately 12% of the respondents in that study reported that their anemia-related fatigue was so severe that they did not wish to continue living. To our knowledge, no studies of the impact of anemia on the quality of life of veterinary cancer patients have been performed. However, it is reasonable to assume that anemia in animals with neoplasia has a similar negative impact on their quality of life. The study of this report was undertaken to assess the impact of anemia on treatment outcome and survival time of dogs with lymphoma that were receiving chemotherapy. Overall, anemia was found to be a risk factor for poor response and short survival time in dogs with lymphoma that were undergoing chemotherapy. Although anemic dogs with lymphoma are not deliberately treated differently from nonanemic dogs with lymphoma at our institute, there was a significant difference in distribution of type of chemotherapy protocol between the 2 groups. Results of the subset analysis of dogs with multicentric lymphoma that were matched on the basis of WHO clinical stage and chemotherapy protocol provided further evidence that anemia is a risk factor for poor response and short survival time.

Among the 96 dogs with lymphoma included in the present study, 40 (42%) were anemic at the time of diagnosis. This is a higher prevalence than that previously reported for tumor-bearing animals² but lower than that reported for humans.¹ Although the precise cause for anemia in the study dogs with lymphoma was not established, it may have been attributable to an inadequate response of the bone marrow to erythropoietin.⁵ Concurrent disease was not detected in the anemic dogs, which suggests that lymphoma was the underlying cause for anemia. Interestingly, the distribution of dogs with cytologic evidence of neoplastic infiltration of bone marrow was similar among the anemic and nonanemic dogs, and the mean Hct value was similar in dogs with WHO stage V lymphoma that did or did not have evidence of bone marrow infiltration. Although acquisition of bone marrow aspirate specimens from multiple sites, collection of bone marrow core biopsy specimens, or performance of molecular or flow cytometric evaluation of bone marrow specimens would have allowed for more sophisticated clinical staging of the neoplasia and may have identified more dogs with WHO stage V disease, it was not possible to apply these techniques because of the retrospective nature of the study.

The dogs with lymphoma in the present study were similar to most descriptions of dogs with lymphoma with respect to age, sex, and predominance of multicentric disease.³⁰ Other prognostic factors that have been variably associated with survival time in dogs with lymphoma, including WHO clinical stage and substage of neoplasia, anatomic form of the lymphoma, and chemotherapy protocol,³⁰ were not significantly associated with outcome in the dogs of the present report. Because of the retrospective nature of our study, it was not possible to gain access to all original biopsy specimens for immunohistochemical evaluation and uniform tumor classification; thus, it was not possible to determine the relative prognostic importance of immunophenotype and histologic characteristics of lymphoma on the outcome in this population of dogs.

The mechanism by which anemia may contribute to decreased response to treatment and reduced survival time is unclear. Chronic hypoxia can induce the expression of vascular endothelial growth factor in a variety of tissues, and increased serum concentrations of this factor have been associated with anemia and poor outcome in humans with cancer, including those with non-Hodgkin's lymphoma.^31,32 Similarly, high serum concentration of vascular endothelial growth factor is a negative prognostic indicator in dogs with lymphoma.³³ Anemia may contribute to tumor hypoxia; tumor hypoxia has an effect on the degree of differentiation of solid tumors, favoring dedifferentiation and a more aggressive phenotype.³⁴ Hypoxia also contributes to the decreased efficacy of ionizing radiation and some chemotherapeutic drugs and plays a role in the development of acquired drug resistance.³⁵ Host hypoxia may have contributed to the poor outcome in dogs with anemia in the study of this report, and may have resulted in the short duration of initial remission in other anemic dogs with lymphoma that were treated with chemotherapy and half-body irradiation.¹⁶

In humans with cancer, treatment with recombinant erythropoietin increases hemoglobin concentration, decreases the need for transfusions, and improves patient-reported quality of life.³⁶ It is also thought that human recombinant erythropoietin is beneficial because it helps to downregulate the hypoxia-inducible factor 1 pathway, which is important for tumor growth.³⁷

The results of the present study suggested that prospective evaluation of anemia management in dogs with lymphoma is warranted. Veterinarians should also be aware that anemia might negatively affect the prognosis of dogs with lymphoma that are undergoing chemotherapy.