Materials and Methods

Dogs—Dogs evaluated at any of the participating institutions between March 1996 and November 2000 in which a diagnosis of malignant lymphoma had been confirmed on the basis of results of histologic or cytologic evaluation were eligible for enrollment in the study. In addition, dogs were included only if they were free from any other serious medical conditions that could have limited full compliance with the study protocol and the owner provided signed consent.

Pretreatment evaluation—All dogs underwent a complete pretreatment evaluation consisting of a physical examination, CBC, serum biochemical panel, urinalysis, thoracic and abdominal radiography, abdominal ultrasonography, and electrocardiography. Echocardiography was performed if indicated on the basis of abnormalities detected during thoracic auscultation, radiography, or electrocardiography. A bone marrow aspirate was obtained and submitted for cytologic evaluation if indicated on the basis of results of the CBC. Tumors were measured directly with calipers or on radiographic or ultrasonographic images. Clinical staging was performed in accordance with the World Health Organization's clinical staging system for lymphoma in domestic animals.¹⁷ Immunophenotyping was performed in a subset of patients by means of immunohistochemical staining with CD3 and CD79a markers or by means of fluorescent-activated cell-sorting analysis^a of a lymph node aspirate.¹⁸

Treatment and monitoring—Dogs were treated, on the basis of owner preference, with a continuous, multiagent chemotherapeutic protocol incorporating L-asparaginase, vincristine, cyclophosphamide, doxorubicin, methotrexate, and prednisolone (Appendix) or with a short-term single-agent protocol that involved administration of doxorubicin (30 mg/m² given as an IV infusion over 30 minutes) every 3 weeks for 5 consecutive treatments. For dogs in the single-agent treatment group, a CBC was performed prior to each doxorubicin treatment. For dogs in both treatment groups, dexamethasone (0.5 mg/kg [0.23 mg/lb], IV) was given before administration of L-asparaginase or doxorubicin, and electrocardiography, along with echocardiography if indicated on the basis of electrocardiographic or auscultatory abnormalities, was performed prior to each doxorubicin treatment. Owners of dogs in the multiagent treatment group were instructed to administer furosemide for 2 days and to frequently walk their dogs after administration of cyclophosphamide.

Serum biochemical analyses were performed when dogs developed any signs of toxicosis or other signs of concern to the attending clinician. Tumor size was evaluated at each visit by direct measurement or measurement on radiographic or ultrasonographic images. In the case of a relapse, patients in the multiagent treatment group were given a second cycle of the induction protocol and patients in the single-agent treatment group were treated with the multiagent protocol. Follow-up examinations following completion of the initial treatment were scheduled monthly during the first 3 months, every 6 weeks for the subsequent 3 months, and every 3 months thereafter.

Response assessment—In evaluating response to treatment, complete remission was defined as a 100% reduction in size of all measurable disease; partial remission was defined as a > 50% but < 100% reduction in size of all measurable disease; stable disease was defined as a < 50% reduction in size of all measurable disease, no change in size, or < 25% increase in size of all measurable disease; and progressive disease was defined as a > 25% increase in size of all measurable disease or the appearance of new lesions. All responses were required to last for at least 14 days. Remissions of shorter duration were classified as stable disease.

Assessment of toxicoses—Hematologic and gastrointestinal tract toxicoses were evaluated 1 week after each treatment. Neutropenia was graded according to criteria established by the Veterinary Co-Operative Oncology Group.¹⁹ Gastrointestinal tract toxicoses were graded from 0 to 4 (grade 0 = no evidence of gastrointestinal tract toxicosis; grade 1 = inappetence, anorexia, vomiting, or diarrhea that was transient or responsive to dietary management; grade 2 = inappetence, vomiting, or diarrhea necessitating medical treatment; grade 3 = inappetence, vomiting, or diarrhea necessitating medical treatment and postponement of the subsequent chemotherapy treatment; and grade 4 = inappetence, vomiting, or diarrhea necessitating medical treatment, postponement of the subsequent chemotherapy treatment, and hospitalization) as described.¹⁵ Other signs of toxicosis such as hemorrhagic cystitis or fever were recorded.

Statistical analysis—Complete and partial remission rates were defined as number of dogs with a complete or partial remission divided by the total number of dogs treated. Kaplan-Meier product-limit analysis was used for analysis of remission and survival times. A complete event was defined as lymphoma relapse or death, respectively. Dogs were censored if they were alive at the time of data analysis (survival analysis) and were still in remission at the time of data analysis or death (remission analysis).

The Kaplan-Meier log-rank test was used to compare remission and survival times between the 2 treatment groups. The Fisher exact test was used to compare remission rates between the 2 treatment groups. The Mann-Whitney U test was used to compare age and body weight between treatment groups. The Fisher exact test was used to compare distributions for sex, clinical stage, substage, anatomic classification, and immunophenotype between treatment groups.

Multivariate Cox forward regression analysis was used to evaluate the following variables for their independent association with remission and survival times: sex, body weight, age, clinical stage, substage, anatomic classification (multicentric vs other), corticosteroid pretreatment, immunophenotype, treatment group, serum creatinine concentration at the time of diagnosis (ie, serum creatinine concentration greater than the upper reference limit of 1.2 mg/dL), high BUN concentration at the time of diagnosis (BUN concentration > 50 mg/dL), hypercalcemia at the time of diagnosis (serum calcium concentration > 2.87 mmol/L or serum ionized calcium concentration > 1.47 mmol/L), anemia at the time of diagnosis (Hct < 40%), and thrombocytopenia at the time of diagnosis (platelet count < 150,000 platelets/ML). Multivariate logistic regression was used to determine whether these factors were associated with the likelihood that dogs would have a complete remission.

All statistical analyses were performed with standard software.^b Values of P < 0.05 were considered significant.

Results

Patient population—A total of 114 dogs were enrolled in the study. Of these, 24 were of mixed breeding, and the remaining 90 represented 37 breeds. Breeds represented by ≥ 4 individuals included German Shepherd Dog (n = 11), Bernese Mountain Dog (10), Rottweiler (8), Airedale Terrier (5), Münsterländer (5), Doberman Pinscher (4), and Schnauzer (4).

Median age was 7 years (range, 1 to 14 years), and median body weight was 32 kg (70.4 lb; range, 4 to 69 kg [8.8 to 151.8 lb]). There were 54 females (24 spayed) and 60 males (12 castrated). Seventy-six of the 114 (67%) dogs had not received any previous treatment. Thirty-eight (33%) dogs had previously been treated with corticosteroids.

Clinical staging—Three of the 114 (2.6%) dogs were classified as stage 1, 3 (2.6%) were classified as stage 2, 38 (33.3%) were classified as stage 3, 37 (32.5%) were classified as stage 4, and 33 (29%) were classified as stage 5. Fifty (43.9%) dogs were substage a, and 64 (56.2%) were substage b. Sixteen of the 38 (42%) dogs with stage 3 lymphoma, 22 of the 37 (59%) dogs with stage 4 lymphoma, and 24 of the 33 (73%) dogs with stage 5 lymphoma were classified as substage b. One hundred six of the 114 (93%) dogs had multicentric lymphoma, 5 (4.4%) had intestinal lymphoma, and 1 each had mediastinal, cutaneous, and nasal lymphoma. Bone marrow aspirates from 64 dogs were submitted for cytologic examination, and 19 of the 64 (30%) had evidence of bone marrow involvement.

Laboratory findings—Thirty-three of the 114 (29%) dogs had anemia at the time of initial examination, and 34 (30%) had thrombocytopenia. Fourteen (12.3%) dogs had hypercalcemia, and 22 (19%) had a high BUN concentration. Fourteen of the dogs with high BUN concentration were treated with the multiagent protocol, and 8 were treated with the single-agent protocol.

Immunophenotyping—Immunophenotyping was performed in 53 dogs. Of these, 44 (83%) had the B-cell phenotype and 9 (17%) had the T-cell phenotype. Of the 9 dogs with T-cell lymphoma, 7 had hypercalcemia and 2 had normocalcemia.

Comparison of treatment groups—Eighty-seven (76%) of the dogs were treated with the multiagent protocol, and 27 (24%) were treated with the single-agent protocol. Median body weight of dogs in the multiagent treatment group (30.5 kg [67.1 lb]; range, 4 to 69 kg [8.8 to 151.8 lb]) was not significantly (P = 0.068) different from median body weight of dogs in the single-agent treatment group (34 kg [74.8 lb]; range, 6 to 45 kg [13.2 to 99 lb]). However, dogs in the multiagent treatment group were significantly (P = 0.041) older (median, 7 years; range, 2 to 14 years) than dogs in the single-agent treatment group (median, 5 years; range, 1 to 13 years). Distributions for sex, clinical stage, substage, anatomic classification, and immunophenotype were not significantly different between groups (Table 1).

Treatment response—One dog treated with the multiagent protocol was lost to follow-up following induction chemotherapy, and no information on treatment response was available. Therefore, this dog was excluded from further analyses.

Of the 86 dogs treated with the multiagent protocol, 63 (73%) had a complete remission, 15 (17%) had a partial remission, and 8 (9%) had stable or progressive disease (ie, no response). Of the 27 dogs treated with the single-agent protocol, 14 (52%) had a complete remission, 6 (22%) had a partial remission, and 7 (26%) had stable or progressive disease. Response distribution was not significantly (P = 0.06) different between treatment groups.

Of the variables examined in the multivariate regression analysis, only substage (P = 0.005) and high BUN concentration (P = 0.004) had significant independent associations with the likelihood that dogs would have a complete remission, with dogs classified as substage ≤ and dogs with a high BUN concentration at the time of initial diagnosis less likely to have a complete remission.

Response after relapse—Thirty-three dogs (27 in the multiagent treatment group and 6 in the single-agent treatment group) that had a relapse received a second round of chemotherapy consisting of the multiagent chemotherapeutic protocol. Response information was available for 31 of the 33 dogs, of which 14 (45%) had a complete remission, 9 (29%) had a partial remission, and 8 (26%) had no response (ie, stable or progressive disease). There was no significant (P = 0.56) difference in response rates between groups when dogs were grouped on the basis of initial treatment group.

First remission duration—Median follow-up time from initial diagnosis was 150 days (range, 3 to 1,669 days; 95% confidence interval, 124 to 176 days).

For dogs treated with the multiagent protocol that had a complete remission (n = 63), median complete remission time was 204 days (range, 19 to 1,669 days) and 6-month, 1-year, and 2-year complete remission rates were 57%, 39%, and 21%, respectively. For dogs treated with the single-agent protocol that had a complete remission (n = 14), median complete remission time was 309 days (range, 105 to 475 days), and 6-month and 1-year complete remission rates were 77% and 19%, respectively. We did not detect a significant (P = 0.63) difference between groups in regard to complete remission time (Figure 1). In this subset of 77 dogs that had a complete remission, Cox regression analysis did not identify any patient variables significantly associated with duration of complete remission.

View Full Size

Kaplan-Meier curves of remission times for dogs with lymphoma that had a complete remission following treatment with a continuous, multiagent chemotherapeutic protocol (n = 63; solid line) or a single-agent protocol involving doxorubicin (14; dotted line). Groups were not significantly (P = 0.63) different.

Citation: Journal of the American Veterinary Medical Association 232, 6; 10.2460/javma.232.6.879

• Download Figure
• Download figure as PowerPoint slide

For dogs treated with the multiagent protocol that had a partial remission (n = 15), median partial remission time was 142 days, and for dogs treated with the single-agent protocol that had a partial remission (6), median partial remission duration was 158 days. No significant (P = 0.84) difference was found between these groups.

Survival—Median overall survival time for dogs treated with the multiagent protocol (n = 86) was 251 days (range, 3 to 1,669 days), and 6-month, 1-year, and 2-year survival rates were 63%, 35%, and 20%, respectively. Median overall survival time for dogs treated with the single-agent protocol (n = 27) was 322 days (range, 5 to 475 days), and 6-month, 1-year, and 2-year survival rates were 60%, 20%, and 0%, respectively. We did not detect a significant (P = 0.38) difference between treatment groups with regard to overall survival time. In the multivariate analysis, substage (P = 0.012) and immunophenotype (P = 0.02) were significantly associated with overall survival time, with dogs with substage a and dogs with the B-cell phenotype having longer survival times.

Median survival time for the 77 dogs with a complete remission was 352 days (range, 19 to 1,669 days). For dogs treated with the multiagent protocol that had a complete remission (n = 63), median survival time was 322 days (range, 19 to 1,669 days), and 6-month, 1-year, and 2-year survival rates were 73%, 43%, and 26%, respectively. For dogs treated with the single-agent protocol that had a complete remission (n = 14), median survival time was 352 days (range, 91 to 475 days), and 6-month, 1-year, and 2-year survival rates were 85%, 31%, and 0%, respectively. We did not detect a significant (P = 0.88) difference between groups in regard to overall survival time (Figure 2). In this subset of dogs, Cox regression analysis did not identify any patient variables significantly associated with survival time.

View Full Size

Kaplan-Meier curves of survival times for dogs with lymphoma that had a complete remission following treatment with a continuous, multiagent chemotherapeutic protocol (n = 63; solid line) or a single-agent protocol involving doxorubicin (14; dotted line). Groups were not significantly (P = 0.88) different.

Citation: Journal of the American Veterinary Medical Association 232, 6; 10.2460/javma.232.6.879

• Download Figure
• Download figure as PowerPoint slide

Median overall survival time for the 21 dogs with a partial remission was 196 days (range, 18 to 476 days). Median survival time for dogs treated with the multiagent protocol (196 days) was not significantly (P = 0.92) different from median survival time for dogs treated with the single-agent protocol (160 days).

Toxicoses—Sixty-nine episodes of neutropenia were documented; 58 episodes occurred in dogs treated with the multiagent protocol, and 11 episodes occurred in dogs treated with the single-agent protocol. Of the 69 episodes of neutropenia, 37 (53%) were grade I, 18 (26%) were grade II, and 14 (20%) were grade III. There was no significant (P = 0.15) difference in the occurrence of neutropenia between the 2 treatment groups.

Fifty-nine episodes of vomiting were documented, of which 52 occurred in dogs treated with the multiagent protocol, and 7 occurred in dogs treated with the single-agent protocol. Of the 59 episodes of vomiting, 27 (49%) were grade 1, 18 (30%) were grade 2, 7 (12%) were grade 3, and 7 (12%) were grade 4. Vomiting occurred significantly (P = 0.022) more frequently in dogs in the multiagent treatment group than in dogs in the single-agent treatment group.

Forty-eight episodes of diarrhea were documented, of which 42 occurred in dogs treated with the multiagent protocol, and 6 occurred in dogs treated with the single-agent protocol. Of the 48 episodes of diarrhea, 23 (48%) were grade 1, 11 (23%) were grade 2, 9 (19%) were grade 3, and 5 (10%) were grade 4. There was no significant (P = 0.051) difference in the occurrence of diarrhea between the 2 treatment groups.

Hemorrhagic cystitis was diagnosed in 8 (9%) dogs, all of which were treated with the multiagent protocol. One dog treated with the multiagent protocol was suspected to have developed neurotoxicosis following vincristine administration. One dog treated with the multiagent protocol died during induction treatment secondary to pyometra, peritonitis, and sepsis.

Discussion

In the present study, we were not able to identify any significant difference in remission or survival times between dogs with lymphoma treated with a continuous, multiagent chemotherapeutic protocol and dogs treated with a short-term single-agent protocol involving doxorubicin. In addition, we were not able to identify any significant difference in the distribution of response rates between treatment groups, although the P value was close to our cutoff for significance. Our results are similar to those of a previous study¹⁰ in which efficacy of doxorubicin administered as a single-agent protocol was similar to efficacy of a combination protocol involving continuous administration of cyclophosphamide, vincristine, and prednisone. Our results may also support results of previous studies^11–15 in which patients receiving short-term chemotherapy had outcomes comparable to those associated with long-term regimens.

A preference in using the combination regime does, however, persist on the authors' side. This subjective preference is partly based on the observation that although no statistical difference could be demonstrated in the present study, dogs receiving the multiagent protocol had 2-year remission and survival probability rates of 21% and 26%, respectively. Additionally, single-agent protocols are in general thought to lead to a more rapid induction of drug resistance. Thus, one could hypothesize that the addition of supplementary drugs to a doxorubicin protocol would result in better treatment outcome.²⁰ Considering these aspects, a multiagent protocol may subjectively be perceived as being advantageous in comparison to a single-agent doxorubicin regime; this, however, remains speculative until future studies demonstrate a statistical difference.

Importantly, in the present study, although we did not find significant differences in median remission and survival times between treatment groups, the short median follow-up time (150 days) and the high rate of censoring may have limited our ability to detect a true difference between groups. Therefore, additional trials with longer follow-up times are needed to verify our findings concerning long-term outcome following treatment and to verify whether the efficacy of the single-agent protocol is indeed equivalent to efficacy of the multiagent protocol.

In the present study, substage and immunophenotype were significantly associated with overall survival time, with dogs with substage a and dogs with the B-cell phenotype having longer survival times. This is in accordance with results of previous studies^4,15,20-24 and reinforces the notion that dogs with lymphoma that have clinical signs or the T-cell phenotype have a poorer prognosis. Interestingly, substage and BUN concentration at the time of initial diagnosis were independently associated with the likelihood that a dog would have a complete remission, with dogs classified as substage ≤ and dogs with a high BUN concentration at the time of initial diagnosis less likely to have a complete remission. Although substage has previously been shown to negatively influence the probability for complete remission,^15,20,21,23 to our knowledge, BUN concentration has not previously been described as an independent prognostic factor for dogs with lymphoma. The reasons for the negative association between BUN concentration and likelihood of a complete remission remain unclear. However, this characteristic may only represent a variation of the finding that dogs with more advanced disease that are sick have a poorer prognosis.¹³

The present study had several limitations that may restrict the validity of our findings. Two of the most important limitations were that dogs were not randomly assigned to treatment groups and that owners and attending clinicians were not blinded to which treatment was being administered. Dogs were assigned to treatment groups on the basis of owner decision, and this may have led to some degree of bias. Many factors can affect an owner's choice of whether to pursue treatment or which treatment to chose, such as perceived prognosis and financial and logistical considerations. This allocation scheme led to an uneven distribution of patient numbers in the 2 treatment groups, which may have limited our ability to detect differences between groups. On the other hand, statistical analysis of patient variables and clinical stage, substage, and immunophenotype did not reveal any differences between the 2 populations. Moreover, diagnostic and staging procedures were identical for dogs in the 2 groups. Only age differed between treatment groups, with dogs treated with the single-agent protocol being slightly younger. This may have exerted a positive influence on remission or survival time for dogs in this treatment group; however, the extent of this remains unclear.

Another important limitation of the present study was that results of immunophenotyping were available for only approximately half the patients. The immunophenotype is of known prognostic importance, with prognosis better for dogs with the B-cell phenotype than for dogs with the T-cell phenotype.^15,20,21 Although there was no difference in phenotype distribution between treatment groups in the present study, the fact that immunophenotype was unknown for a large subset of patients may have introduced some bias.

Finally, it is possible that the number of patients included in the present study was insufficient to demonstrate a significant difference in efficacy between treatment groups. A higher number of cases and an even distribution of cases between groups may have led to the detection of differences in remission and survival times. Retrospective power calculations revealed low power (14%) and showed that 82 patients would have been needed in each group to be 90% certain of detecting a 50% improvement in survival time between groups. To be 90% certain of detecting a 30% improvement in survival time between groups, it would have been necessary to enroll 227 dogs in each treatment group, and it is difficult to accrue these case numbers in veterinary clinical research, even when performing multi-institutional studies.

Interestingly, dogs in the present study treated with the multiagent protocol that had a complete remission had a median remission time of only 204 days, which was substantially less than remission times of 252, 282, and 300 days and 55 weeks reported in previous studies.^5,6,12,14 On the other hand, remission and survival times in the present study were comparable to those reported with other protocols.^{4,7,9,11,22,25-27}

It is possible that the inclusion of dogs with nonmulticentric lymphoma in the present study may have contributed to shorter remission times, even though Cox regression analysis did not reveal any significant association between anatomic classification and outcome. Additionally, the present population may have differed from populations described in North American studies (eg, neuter status, breed distribution, or substage distribution), which may have contributed to differences in response duration. For these reasons, results of comparison between the 2 treatment protocols in the present study must be interpreted carefully. Furthermore, the external validity of a clinical trial is subject to limitations associated with a number of factors, such as selection bias and population differences.^28,29 These limitations must be taken into consideration when extrapolating results of a study group to the general population.^28,29