Reevaluation of the University of Wisconsin 2-Year Protocol for Treating Canine Lymphosarcoma
This retrospective study investigated a population of 96 dogs with newly diagnosed malignant lymphosarcoma that were treated with the commonly used University of Wisconsin-Madison (UW-M) chemotherapy protocol. Pretreatment characteristics were analyzed to determine prognostic factors. Dogs with higher World Health Organization (WHO) stages (including stage IV) and dogs with hypercalcemia were at significantly higher risk of relapse (P=0.018 and P=0.016, respectively). Dose reduction, treatment delays, and prior therapy with cortico-steroids were not associated with clinical outcome. First remission duration of 270 days was similar to historically reported data. Overall survival time of 218 days was much shorter than historical data.
Introduction
Lymphosarcoma, neoplasia arising from lymphoreticular cells, is one of the malignancies most likely to respond to chemotherapy, and it is also one of the most common neoplasms of the dog.1 Various chemotherapeutic regimens have been used and usually consist of multiple drugs.2,3 Response rates of 80% to 90% with median survival times of 250 to 300 days have been reported.2,4 Numerous studies have been undertaken to identify prognostic factors relating to response to treatment, length of remission, and survival times. Results from these studies have varied.5,6–11 One common treatment regimen used for canine lymphosarcoma is the University of Wisconsin-Madison (UW-M) 2-year chemotherapy protocol. Studies with limited numbers of dogs have investigated survival times of dogs treated with this protocol.6,12,13
The objective of this retrospective study was to evaluate signalment, prognosis, and prognostic factors in a previously unreported canine population that underwent treatment with the UW-M chemotherapy protocol for lymphosarcoma.
Materials and Methods
Case Selection
Medical records from April 1996 to December 2002 were searched at the Small Animal Clinic, Veterinary School, University of Zurich, for dogs with definitive cytological or histopathological diagnosis of lymphosarcoma that had received no chemotherapy except corticosteroids before being treated with the UW-M 2-year chemotherapy protocol. Diagnosis of lymphosarcoma was made on the basis of cytological examination of needle aspirates and/or histological examination of surgically excised lymph nodes or other tissue samples. If any question concerning the accuracy of a cytological diagnosis arose, histological examination of a tissue specimen was performed.
Staging was done using the modified World Health Organization (WHO) staging system for canine lymphosarcoma and consisted of a complete blood count (CBC), a serum biochemical profile, urinalysis, and thoracic radiography.2 Whenever the physical examination and/or the serum biochemical profile suggested visceral involvement, abdominal ultrasonography (n=27) or radiography (n=31) was performed. Leukemia was diagnosed based on bone marrow evaluation (n=6) or cytological examination of blood smears (n=11). The WHO staging system was modified as described in a previous study, so that extranodal lymphosarcomas were considered a separate stage.12 Stage VI was created based on the fact that extranodal lymphosarcomas do not clinically behave like stage V lymphosarcomas [Table 1].
The dogs were further divided into substage a (without clinical signs) and substage b (with clinical signs of illness). Factors that were investigated included age, gender, breed, weight, clinical stage and substage, anatomical location, presence of hypercalcemia (reference range 9.6 to 11.2 mg/dL) or hyperbilirubinemia (reference range 0.145 to 0.44 mg/dL), hematocrit, hemoglobin, presence of neurological signs, prior treatments, time between onset of signs until first treatment, toxicities associated with chemotherapy, dose reductions and alterations in therapy, and therapy instituted at relapse. Month of presentation and month of relapse were also noted in order to evaluate for seasonality, a factor anecdotally reported in dogs. Because absolute thrombocyte counts were not available in most cases (i.e., platelet numbers were indicated as sufficient, high, or low), this parameter was not evaluated. Immunophenotyping was not routinely performed, owing to cost concerns.
Treatment Protocol
The doses and frequency of drugs used in the UW-M protocol are shown in Table 2. In 22 dogs, the first dose of vincristineb was reduced to 0.5 mg/m2 because of concerns of increased toxicity in the first week of therapy when L-asparaginase and vincristine were combined.14,15
Side Effects
Toxicity was assessed based on the National Cancer Institute (NCI) toxicity grading system for hematological and gastrointestinal toxicities [Table 3]. Owners were questioned as to whether the dog had any signs of gastrointestinal toxicosis (e.g., anorexia, vomiting, or diarrhea) and whether the dog’s demeanor had changed. In cases of myelosuppression (i.e., a neutrophil count <3000/μL), treatment was delayed for a week, and subsequent doses of that drug were reduced. If severe drug-induced gastrointestinal toxicosis developed, therapy was withheld, and the dosage of the agent thought to be the cause of toxicosis was reduced. Only treatment delays resulting from toxicity were statistically evaluated.
Responses and Outcomes
Complete remission was defined as the disappearance of all clinical evidence of disease on physical examination, radiography, ultrasonography, CBC, and/or biochemical analyses. A partial response was defined as a decrease in tumor volume =50%, with no new lesions. Stable disease was defined as no change in tumor burden or <50% decrease in tumor volume. Progressive disease was defined as appearance of new lesions and/or tumor growth.
The first remission duration was defined as the time between recognition of a complete remission until evidence of relapse. Overall survival time was defined as the time from the first chemotherapy treatment until death from lymphosarcoma or treatment. Survival time was determined by review of the medical record or by contact with the referring veterinarian or owner.
Statistical Analysis
All dogs entering the study were included in analysis for overall survival time (n=96). Dogs that received only one treatment and were euthanized or died were evaluated, despite not completing the induction protocol. Only dogs that achieved remission were included in remission duration calculations (n=76). Discrete factors evaluated for significance included gender, breed, clinical stage and substage, anatomical location, presence of hypercalcemia, presence of neurological signs, prior treatments, toxicities experienced, dose reductions, alterations in therapy, and therapy at relapse. Continuous factors included age, body weight, hyperbilirubinemia, hematocrit, hemoglobin, time between onset of disease until treatment, and toxicities experienced. Univariate analysis to assess the prognostic values of the different discrete cofactors was performed by the Kaplan-Meier method, together with the logrank or “Breslow” test.16 The associations between two discrete covariates were evaluated by Fisher’s exact test, using a computerized software program.a In order to compare means of a continuous variable with respect to a factor with only two levels, the Mann-Whitney U test was applied.16 For two continuous covariates, Pearson’s correlation coefficient was computed with a 95% confidence interval.16 Month of presentation and month of relapse were evaluated by the goodness of fit chi-square (χ2) test.17 Variables that were identified to be significant in univariate analysis were included in a multivariate Cox proportional hazards model regression analysis with a forward and backward step selection for both survival and remission.16 Results were considered to be significant with P values <0.05.
Dogs were censored in analysis of remission duration if they were in remission at the end of the study, were lost to follow-up, or had died from other causes while in remission. Dogs were censored in analysis of survival if they were lost to follow-up, died from causes unrelated to lymphosarcoma or chemotherapy, or were alive at the end of the study period.
Results
Case Data
Ninety-six dogs met the inclusion criteria for the study. The median age at diagnosis was 7.7 years (range 0.5 to 14 years; mean 7.63±0.3 years). Forty-six dogs were females (32 spayed), and 50 were males (18 castrated). The most common breeds included the Bernese mountain dog (n=11), rottweiler (n=5), German shepherd dog (n=5), Doberman pinscher (n=4), and Labrador retriever (n=4). Forty-seven (49%) dogs were composed of 30 additional breeds, and 20 (21%) animals were mixed-breed dogs. The median weight was 31 kg (range 4 to 78 kg; mean 30.0±1.5 kg). Stage distribution is shown in Table 4.
Seventeen (18%) dogs were leukemic, 17 (18%) had cranial mediastinal involvement, and six (6%) had pulmonic involvement. The sites of involvement of the nine stage-VI dogs were gastrointestinal (n=3), ocular (n=3), nasal (n=1), and spinal (n=2). Fifteen (16%) dogs had hypercalcemia (median 14.8 mg/dL, range 12.5 to 18.8 mg/dL), eight of which had mediastinal involvement (i.e., stage I, mediastinum only [n=1]; stage II [n=1]; stage III [n=4]; and stage IV [n=2]). Hypercalcemic dogs without mediastinal involvement were staged as III (n=5) and IV (n=2). Nine dogs had hyperbilirubinemia. The median bilirubin level of all dogs measured was 0.19 mg/dL (range 0.005 to 14.15 mg/dL). The median hematocrit was 42.0%±0.8% (range 18% to 61%), and the median hemoglobin was 14.4±0.3 g/dL (range 3.8 to 21 g/dL). Three dogs were presented with neurological signs. Correlations were found between WHO stage and age (P=0.0002), WHO stage and hypercalcemia (P=0.0227), and WHO stage and age and weight (P=0.021). Hypercalcemia was found in younger dogs in lower stages. Higher WHO stages were associated with greater age and smaller body weights.
Fifty-four dogs received prior treatments that consisted of antibiotics (n=24), corticosteroids (n=14), antibiotics and corticosteroids (n=11), or alternative medications, such as mistletoe or immunomodulatory drugs (n=5). Median duration of pretreatment was 13.5±2.4 days (range 1 to 81 days). Median time from appearance of first clinical signs until treatment was 18±2.8 days (range 4 to 165 days). The initial diagnoses were highest in the months of July (n=10), August (n=10), November (n=15), and December (n=10), but this pattern was not statistically significant (χ2 test=16.05; P=0.1393). A higher tendency of relapse was detected in the months of May (n=8), June (n=7), July (n=7), September (n=9), and October (n=7) (χ2 test=15.75; P=0.1507). In the years from 1996 until the end of the study in 2004, not one dog relapsed in the month of April.
Side Effects
Side effects and dosage reductions are summarized in Tables 5 and 6. Side effects were most frequently experienced after the first week of treatment. The five dogs that died after their first treatment were all substage b in stage III (n=2), stage IV (n=2), or stage VI (n=1). Twelve dogs suffered bone marrow toxicity after week 1, and 26 dogs had gastrointestinal toxicities; some dogs suffered concurrently from both. Sixteen dogs had to be hospitalized (mean duration 2.9±0.7 days, range 1 to 11 days) for drug-induced toxicity. Six (6.2%) hospitalizations occurred after week 1. Eleven dogs were hospitalized once, and five dogs were hospitalized more than once (two to three times). The drug that most frequently caused side effects (in 66 dogs) was vincristine. Thirteen dogs exhibited gastrointestinal toxicity after the first doxorubicin treatment in week 4. In one case, vincristine-induced neuropathy was strongly suspected, and vincristine was discontinued at week 15 of the protocol. Treatment delays owing to toxicity usually occurred during the induction phase (n=39). Remission duration was not significantly associated with dosage reductions (n=43; P=0.248) or treatment delays (n=35; P=0.233). Specifically for the 22 dogs that received vincristine at 0.5 mg/m2 in week 1, there was no statistically significant influence on remission duration (n=19; P=0.10), and toxicities in week 1 for this group of dogs included gastrointestinal (n=6, 27%), bone marrow (n=3, 14%), hospitalizations (n=2, 9%), and death (n=1, 4%).
Treatment Response
Seventy-six (79%) dogs experienced a complete remission. Dogs that did not achieve a complete response had various stages of disease, namely stage II (n=1), stage III (n=6), stage IV (n=6), stage V (n=4), and stage VI (n=3). The median duration of the first remission was 270±20.3 days. Breed and gender status had no statistical influence on remission. Univariate variables that significantly increased the likelihood of relapse were hypercalcemia (P=0.017) and higher initial stage classification (stages IV, V, and VI as opposed to stages I, II, and III; P=0.018).
Of the 15 dogs that were hypercalcemic, 11 had a complete response. The median duration of the first remission for the 11 dogs was only 139±26.3 days, as opposed to 296±31.2 days for dogs that were normocalcemic. Hypercalcemia did not influence the likelihood of achieving remission. In univariate analysis, the presence of leukemia resulted in shorter remission duration. Leukemic dogs had a median remission duration of 172±9 days, as opposed to 296±30.5 days for nonleukemic dogs (P=0.02). A tendency for longer remission duration was seen in dogs classified as substage a (308±43.7 days) versus substage b (176±55.3 days), but substage alone was not statistically significant (P=0.16). Being classified as substage a was not a significant predictor of the likelihood of remission (P=0.21). The time period between onset of the disease and the first chemotherapeutic administration was not associated with achievement or duration of remission (P=0.90). Of the 19 dogs pretreated with corticosteroids, remission duration was not significantly impacted (P=0.70).
Evaluation of multiple Cox regression by the backward search procedure revealed only two factors that remained statistically significant, namely elevated calcium (P=0.037; Figure 1) and higher WHO classification of stage (P=0.016; Figure 2). Forward stepwise analysis revealed leukemia as a negative prognostic marker (P=0.011), together with hypercalcemia (P=0.007).
In 34 dogs, a second remission was attempted with either the same protocol (n=17) or with lomustineh (60 mg/m2 per os q 3 weeks; n=17). A number of dogs received other rescue therapies, such as mitoxantrone (n=1), actinomycin (n=1), melphalan (n=1), or no treatment (n=16), but they were not included in the second remission analysis because of low numbers in each treatment group. Eighteen (53%) of 34 dogs achieved a second remission; 11 of these 18 received the same protocol, and seven received lomustine. A significant difference in the length of second remission duration was detected between those treated with the same protocol and those treated with lomustine; dogs that were reinduced with the same protocol had a significantly longer overall survival time than the latter group (P=0.008; Figure 3).
Outcomes
Three dogs were still in remission at the end of the study (845, 1336, and 1720 days after initial treatment). Eight dogs had died or were euthanized for causes unrelated to lymphosarcoma, with various dogs experiencing the following: cardiomyopathy (week 3 of therapy), ophthalmic disease, gastric foreign body, neurological disease, renal failure, osteosarcoma, orthopedic disease, and sudden death after a dental procedure. Four dogs were lost to follow-up at 329, 456, 806, and 1333 days. The median survival time for all dogs was 218±15.90 days. Univariate analysis identified WHO substage as a significant (P=0.004) prognostic indicator of overall survival time. The median overall survival time for dogs in substage a was 443±136.1 days; the median overall survival time for dogs in substage b was 136±77.3 days. Median overall survival time for dogs that achieved a complete remission was 322±68.6 days.
Discussion
Many findings of this study were similar to previously published results. The median age and weight of the dogs were similar to other studies.12,13 The high number of intact females in this study may have been related to regional differences, because in North American studies, intact females are typically underrepresented.9,14,18 In the present study, clinical courses experienced by intact females were similar to courses for spayed females and males. One new finding was the high percentage of Bernese mountain dogs, which might indicate a susceptibility to lymphosarcoma in this breed. A comparison of affected dogs with the overall hospital population was not performed, however.
The current study identified hypercalcemia as a significant prognostic factor in univariate and multivariate analyses. This finding has been reported previously using univariate analysis in some studies, whereas other studies have reported no significance of hypercalcemia as a prognostic indicator.9,12,14,19 The study reported here further identified a negative correlation between hypercalcemia and age and a positive association between hypercalcemia and lower WHO stages (a finding not previously reported). One European study found a WHO classification of stage IV to be a positive prognostic indicator for complete remission rates, whereas the present study identified stages IV, V, and VI as negative prognostic indicators.2
Dogs that received prednisone before starting the protocol were not at a disadvantage; this finding was similar to a former study.9 Other studies have reported that delays in therapy and dosage adjustments were most frequently associated with vincristine, a finding that was confirmed in this study.14 Lowering the vincristine dose at week 1 did appear to decrease toxicity, because only 14% of the dogs in this study experienced bone marrow toxicity, versus 40% in a study by Northrup et al.15 In the present study, significant toxicity did occur in dogs given a reduced dose of vincristine in week 1, at a level similar to previous reports on this protocol. This finding supported the contention that it is the combination of vincristine and L-asparaginase, more than the dose of vincristine alone, that causes bone marrow suppression.14,19 Evaluation of vincristine toxicity in week 1 of the present study was skewed, however, in that certain dogs (often those with substage b or higher tumor burdens and considered to be at greater risk for toxicity) were selected to receive a reduced dose of vincristine at week 1. Avoiding toxicity at week 1 allowed subsequent administration of full dosages of vincristine for the remainder of the protocol (i.e., a permanent dose reduction was not made, based on the dog having had a toxic reaction to the vincristine at week 1). Remission duration for dogs that received vincristine at 0.5 mg/m2 in week 1 was not shorter than the remission duration for the group that had the full dose, so it may be useful to consider a dose reduction in all dogs at week 1, or at least in those that are considered to be at the highest risk for side effects during initial induction.
The complete response rate of 79% was lower in this study than the rate previously reported (84% to 87%) using the same protocol.12,13 One European study also observed a similar 77% complete response rate.2 The median duration of the first remission was comparable to other studies.8,12–14 The overall survival time was lower at 218 days than the previously reported survival time of 357 days.12 The lower survival time may have been a real survival disadvantage, or it may have arisen from cultural differences, with European clients less likely to try multiple rescue protocols or choosing euthanasia earlier.
The current study contained several limitations; besides its retrospective nature, the study had a lack of control dogs, and differences existed in diagnostic testing procedures, ancillary treatments, and methods of evaluation and follow-up. The large number of dogs in this study helped to minimize some of these problems. Another important point was the possibility of some degree of backward stage migration, as low numbers of dogs placed in stages I to III (n=43) had abdominal ultrasonography (n=7) or radiography (n=10) performed. As many as 26 dogs may have been staged lower than they actually were. When these 26 dogs were experimentally classified as stage IV to VI, there was no longer a statistically significant difference between dogs in stages I to III (n=17) and dogs in stages IV to VI (n=79; P=0.32).
Additional factors may also have influenced treatment outcomes in the dogs of this study. Immunophenotyping has been used to identify prognostic differences between the cellular subtypes of canine lymphosarcoma, but it was not performed in the study reported here.21
Conclusion
Medical records for 96 dogs with lymphosarcoma treated with the UW-M 2-year chemotherapeutic protocol were reviewed. Hypercalcemia and higher stage of disease were confirmed as negative prognostic indicators independent of other variables. Treatment with corticosteroids prior to starting the chemotherapy protocol did not affect remission or survival. A correlation was found between young dogs, hypercalcemia, and low-stage disease. A trend was seen for dogs being presented more often in winter (November and December) and summer (July and August). Dose reduction of vincristine in week 1 appeared to decrease toxicity without affecting remission length or survival, and this should be considered as a viable modification of the protocol, particularly for dogs that are substage b.
Statview version 5.0.1; SAS Institute, Inc., Cary, NC 27513
Oncovi; Medika AG, Switzerland
Leunase; Farmamondo, Chiasso, Switzerland
Endoxan; Baxter AG, Switzerland
Adriablastin RD; Pfizer AG, Zürich, Switzerland
Leukeran; GlaxoSmithKline, Muenchenbuchsee, Switzerland
Methotrexat Proreo; Proreo Pharma AG, Liestal, Switzerland
Prava; Bristol-Myers Squibb GmbH, Baar, Switzerland
Citation: Journal of the American Animal Hospital Association 43, 2; 10.5326/0430085
Citation: Journal of the American Animal Hospital Association 43, 2; 10.5326/0430085
Citation: Journal of the American Animal Hospital Association 43, 2; 10.5326/0430085
Kaplan-Meier median remission duration data comparing hypercalcemic and normocalcemic dogs with lymphoma that were treated with the University of Wisconsin-Madison 2-year chemotherapy protocol. Median remission durations were 139±20.3 days and 296±31.2 days, respectively (P=0.017).
Kaplan-Meier median remission duration data comparing dogs with lymphosarcoma classified as stages I to III and dogs with lymphosarcoma classified as stages IV to VI (according to the World Health Organization Clinical Staging). Median remission durations were 327±9.5 days and 207±83.2 days, respectively (P=0.018).
Kaplan-Meier median second remission duration data comparing dogs treated with lomustine (60 mg/m2 per os q 3 weeks) to dogs treated with the University of Wisconsin-Madison 2-year chemotherapy protocol at relapse. Median second remission durations were 29±15.1 days and 60±8.7 days, respectively (P=0.008).
Contributor Notes
