Vincristine is among the anticancer drugs most commonly used for treatment of dogs with lymphoma.1–4 Adverse effects associated with vincristine in dogs with lymphoma include gastrointestinal toxicosis and hematologic toxicosis (myelosuppression).3–5 Hematologic toxicosis affected 23 of 40 (58%) dogs with lymphoma treated with vincristine (initial dose, 0.7 mg/m2) in 1 study3 and 12 of 96 (13%) similar dogs treated with vincristine (initial dose, 0.5 mg/m2) in another study.4 Hematologic toxicosis, primarily neutropenia and thrombocytopenia, may result in treatment delays or required dose reductions. These are undesirable because they may reduce the likelihood of achieving remission, reduce owner compliance to continue treatment, or shorten remission duration. Complications such as sepsis are associated with neutropenia and can increase morbidity and mortality rates, providing further incentive for avoiding hematologic toxicosis.5
In humans, a 10-fold variability in vincristine disposition has been reported, and much of that variability has been attributed to pharmacogenetic polymorphisms.6 Pharmacogenetic polymorphisms also exist in dogs and can contribute to individual as well as breed-related variations in drug disposition. A breed-related polymorphism in ABCB1 (ie, ABCB1-1Δ) increases the risk of vincristine-associated hematologic toxicosis in dogs.7 Breeds known to carry the ABCB1-1Δ mutation include several herding breeds (including Border Collies) and 2 breeds of sight hounds. Dogs heterozygous or homozygous for the ABCB1-1Δ mutation are significantly more likely to develop hematologic toxicosis, specifically neutropenia and thrombocytopenia, after treatment with vincristine than are ABCB1-wildtype dogs.7 In that study,7 6 of 8 dogs with the ABCB1-1Δ mutation had neutropenia and 5 of 8 had thrombocytopenia, compared with only 3 of 26 and 1 of 26, respectively, ABCB1-wildtype dogs.
Because a single polymorphism is not likely to account for variability in vincristine disposition for all dogs, it is reasonable to expect that some ABCB1-wildtype/wildtype dogs will develop myelosuppression associated with vincristine administration. Indeed, 3 of 26 ABCB1-wildtype/wildtype dogs in the aforementioned study7 developed myelosuppression. Although that study7 population was biased toward dogs of herding breeds, it was surprising that the only 3 ABCB1-wild-type dogs that developed myelosuppression were Border Collies. On the basis of that observation, it seemed reasonable to hypothesize that Border Collies are predisposed to VAM and that this breed-related sensitivity is mediated by a different ABCB1 polymorphism.
The gene that encodes P-glycoprotein is ABCB1 (formerly known as multidrug resistance 1 gene). P-glycoprotein is a membrane transporter that actively pumps substrate drugs into bile and urine for excretion from the body. It also serves as an important component of the blood-brain barrier by transporting drugs back into the capillary lumen from within capillary endothelial cells in the brain. Drugs known to be substrates for transport by P-glycoprotein include antiparasitic drugs (avermectins), loperamide, and many anticancer agents, including vincristine. Inhibition of P-glycoprotein function in rats causes a 4-fold decrease in biliary clearance of vincristine and a 30% decrease in its renal clearance.8 Because of the importance of P-glycoprotein in biliary and renal clearance of vincristine, we hypothesized that affected Border Collies would have a functional polymorphism in ABCB1 distinct from the previously characterized ABCB1-1Δ polymorphism. Therefore, the objective of the study reported here was to determine whether Border Collies (ABCB1 wildtype) were more likely than other breeds to develop VAM and, if so, whether this was caused by a mutation in ABCB1 distinct from ABCB1-1Δ.
Materials and Methods
Animals—Phase 1 of the study included 26 ABCB1-wildtype dogs with lymphoma (5 Border Collies and 21 dogs of other breeds) treated with vincristine that had been enrolled in a previous prospective study.7 Phase 2 of the study included 10 additional Border Collies. The study was approved by an institutional animal care and use committee.
For phase 1, collection of DNA from dogs was accomplished by soliciting participation via veterinarians (veterinary oncologists) and owners. Samples were submitted from the Washington State University College of Veterinary Medicine; Veterinary Oncology Services PLLC, Hopewell Junction, NY; Red Bank Veterinary Hospital, Tinton Falls, NJ; The Animal Medical Center, New York; and the North Carolina State University College of Veterinary Medicine.
The previous study7 consisted of 34 dogs receiving chemotherapy with any vincristine-containing protocol for treatment of lymphoma. Eight dogs with positive results when tested for the ABCB1-1Δ mutation were excluded from the present study, but the remaining 26 ABCB1-wildtype dogs were included. Breed of each dog and the dosage of vincristine administered were recorded. Accessions in that previous study7 initially included dogs of all breeds to ensure an adequate population of ABCB1-1Δ–wildtype dogs, but accessions subsequently were limited to only herding breeds. Dogs were treated on the basis of each facility's standard treatment practices for dogs with lymphoma, with hematologic adverse reactions attributed to vincristine administration graded in accordance with the VCOG criteria for adverse event reporting,9 with 1 modification (ie, a grade of 0 was assigned to categories in which the patient did not have thrombocytopenia or neutropenia). Affected dogs were those that had hematologic toxicosis (ie, received a VCOG grade ≥ 1 for neutropenia or thrombocytopenia), whereas unaffected dogs were those that received chemotherapy for lymphoma via a vincristine-containing protocol but did not develop hematologic toxicosis (ie, received a VCOG grade of 0 for neutropenia and thrombocytopenia). Dogs that had received a chemotherapeutic drug, except for a corticosteroid, concurrently with vincristine were excluded from the study, with 1 exception. If a dog received l-asparaginase concurrently with vincristine, and that dog did not have an adverse event, then that dog was included in the study. A CBC conducted on a blood sample obtained between 5 and 15 days after vincristine treatment was used to determine vincristine-associated hematologic toxicosis. For one of the affected Border Collies, a CBC was not obtained until 18 days after treatment with vincristine; however, because of extenuating circumstances, the dog had not received any drug treatment after administration of the vincristine and prior to the CBC. Thus, myelosuppression in that dog was ascribed to vincristine and the dog was included in the study.
Values for neutrophils and thrombocytes, including abnormal values, were included in the statistical analysis. Data for hemoglobin concentration and PCV were not analyzed in the present study. Veterinary oncologists were not provided information regarding the ABCB1 genotype of a dog until the VCOG grade for that dog had been submitted. Furthermore, the investigator performing the ABCB1 genotyping was not informed of the VCOG grade for the dogs until the end of the study.
For phase 2 of the study, DNA from 10 Border Collies was used for ABCB1 sequencing. Three of these Border Collies were identified through a genotyping laboratorya on the basis that they had developed hematologic toxicosis after treatment with vincristine (despite having the ABCB1-wildtype genotype) and had not received another antineoplastic agent other than prednisone. The other 7 Border Collies were of unknown phenotype and had not been treated with vincristine.
ABCB1 sequencing and sequence analysis—The PCR primers were designed to amplify all 27 ABCB1 exons from genomic DNA (Appendix). Primers were created with a program for designing PCR primersb and were intended to anneal in the flanking intron sequences. Nucleotide sequences were evaluated to detect differences between affected dogs, the published canine ABCB1 sequence, unaffected dogs, and dogs of unknown phenotype that had not been treated with vincristine. A base pair change was considered to be potentially causative for vincristine-associated hematologic toxicosis when it met 3 criteria: it was identified in at least 2 of the affected dogs, it was not identified in any of the unaffected dogs or the published canine ABCB1 sequence, and it was a nonsynonymous change that resulted in an amino acid of different polarity or charge. A base pair change that met those criteria was further evaluated with a polymorphism phenotyping toolc to assess the amino acid change on predicted protein structure or function.
Statistical analysis—Data for dogs solely from phase 1 (recruited prospectively with a blinded endpoint) were included in the statistical analysis to determine whether Border Collies were more likely than other breeds to develop VAM. Because several cells had small expected values, a Fisher exact test was used to evaluate whether the proportion of dogs that developed VAM differed between Border Collies and dogs of other breeds.
Results
Border Collies were significantly (P = 0.024; Fisher exact test) more likely to develop vincristine-associated hematologic toxicosis (3/5) than were any other breed of dog (0/21). There was not a significant (P = 0.77) difference in the mean ± SD dose of vincristine (0.60 ± 0.10 mg/m2) administered to Border Collies, compared with the dose (0.63 ± 0.09 mg/m2) administered to other breeds. The proportion of Border Collies and other breeds that developed hematologic toxicosis was summarized (Table 1).
Breeds of ABCB1-wildtype dogs treated with vincristine and the proportion of those dogs that developed hematologic toxicosis.
| Breed | Vincristine dose (mg/m2) | Proportion of dogs with hematologic toxicosis |
|---|---|---|
| Border Collie (n = 5) | 0.5 (n = 2); 0.6 (n = 1); 0.7 (n = 2) | 3/5* |
| Collie (n = 2) | 0.5 (n = 2) | 0/2 |
| Labrador Retriever (n = 3) | 0.5 (n = 1); 0.7 (n = 2) | 0/3 |
| Shetland Sheepdog (n = 4) | 0.5 (n = 1); 0.6 (n = 1); 0.7 (n = 2) | 0/4 |
| Australian Shepherd (n = 3) | 0.6 (n = 1); 0.7 (n = 2) | 0/3 |
| Other breeds (n = 9)† | 0.5 (n = 3); 0.6 (n = 2); 0.7 (n = 4) | 0/9 |
The vincristine dose administered to the 3 dogs with hematologic toxicosis was 0.5, 0.6, and 0.7 mg/m2, respectively.
Other breeds are Rottweiler, Cocker Spaniel, mixed, Welsh Corgi, Miniature Schnauzer, German Shepherd Dog, Staffordshire Terrier, Australian Cattle Dog, and Siberian Husky.
For all 15 Border Collies included in both phases of the present study, the entire ABCB1 coding region (3,842 bp), which includes 27 exons, was sequenced. Eight SNPs involving 7 exons were identified (Table 2). Because no SNPs segregated with the affected phenotype, none of the SNPs met the criteria as potentially causative for VAM. Exon 10 contained 2 apparently independent SNPs. All of the identified SNPs were present at a low frequency (present in the heterozygote state in only 1 or 2 of the 15 Border Collies). No insertion or deletion mutations were detected.
The SNPs found in ABCB1 exon sequences of 15 Border Collies.
| Nucleotide | Reference sequence | With VAHT | Without VAHT | Unknown VAHT status | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A1 | A2 | A3 | A4 | A5 | A6 | C1 | C2 | U1 | U2 | U3 | U4 | U5 | U6 | U7 | SNP details | ||
| 504 | C* | Y* | C | C | C | C | C | C | C | C | — | — | C | C | C | C | exon5, Asp168Asp |
| 1,149 | A | A | R* | A | A | A | A | A | A | A | — | A | A | A | A | A | exon10, Gly383Gly |
| 1,186 | A | A | A | A | A | A | A | R* | A | A | — | A | A | A | A | A | exon10, Lys396Glu |
| 1,480 | G | G | G | G | G | G | G | R* | G | G | — | G | G | G | G | G | exon12, Glu494Lys |
| 1,663 | C* | Y* | C | C | — | C | C | C | C | C | C | C | C | Y* | C | C | exon13, Leu555Leu |
| 3,017 | C | C | C | C | C | C | C | C | C | C | Y* | C | C | C | C | C | exon23, Ala1006Val |
| 3,439 | A | A | A | A | R* | A | A | A | A | A | A | A | A | R* | A | A | exon25, Met1147Val |
| 3,567 | C | C | C | C | C | C | C | Y* | C | C | C | Y* | C | C | C | C | exon26, Arg1189Arg |
| Vincristine dose (mg/m2) | NA | 0.5 | 0.6 | 0.7 | 0.5 | 0.7 | 0.6 | 0.7 | 0.5 | — | — | — | — | — | — | — | |
| VCOG neutropenia grade | NA | 2 | 3 | 1 | 1 | 3 | 3 | 0 | 0 | — | — | — | — | — | — | — | |
| VCOG thrombocytopenia grade | NA | 2 | 0 | 0 | 0 | — | 0 | 0 | 0 | — | — | — | — | — | — | — | |
Letters indicate nucleotides present at SNP site. Reference sequence NM_001003215.1 (Beagle) was used as the basis for numbering of bases and amino acids.
Dog was heterozygous for more frequent allele; all other dogs for which data were available were homozygous.
— = Not determined. A = Affected. C = Control. NA = Not applicable. R = A/G heterozygote. U = Unknown phenotype. VAHT = Vincristine-associated hematologic toxicosis. Y = C/T heterozygote.
Four of the SNPs were synonymous (SNPs involving exons 5, 10, 13, and 26), and 4 were nonsynonymous (SNPs involving exons 10, 12, 23, and 25). The nonsynonymous SNPs resulted in the following amino acid changes: exon 10, lysine (hydrophilic and positively charged) to glutamate (hydrophilic and negatively charged); exon 12, glutamate (hydrophilic and negatively charged) to lysine (hydrophilic and positively charged); exon 23, alanine (hydrophobic) to valine (hydrophobic); and exon 25, valine (hydrophobic) to methionine (hydrophobic).
Discussion
In a previous study,7 dogs that had 1 or 2 copies of the ABCB1-1Δ mutation were found to be significantly more likely to have VAM than dogs that did not have the ABCB1-1Δ mutation. In that study,7 the only dogs that had vincristine-associated hematologic toxicosis and did not have the ABCB1-1Δ mutation were Border Collies, although the sample population was biased toward dogs of herding breeds. In the present study, we did not find any coding mutations in the ABCB1 gene that would account for VAM in these Border Collies. However, ABCB1 cannot be definitively excluded as a candidate gene for VAM. It is possible that mutations involving the ABCB1 promoter, intronic regions, or epigenetic changes could affect P-glycoprotein expression or function in these patients.
An insertion mutation was recently described10 for a Border Collie that developed signs of ivermectin toxicosis after receiving an unknown dose of ivermectin.10 None of the siblings or parents of that dog had signs of ivermectin toxicosis after exposure, but the doses those dogs received also were unknown. The ABCB1 cDNA isolated from that ivermectin-sensitive Border Collie as well as from both parents and 5 siblings (1 from the same litter and 4 from 3 other litters) was sequenced. Several polymorphisms in the ABCB1 cDNA sequence were identified, but the only 1 that was unique to the ivermectin-sensitive dog was a 3-bp insertion.10 That insertion results in the addition of an asparagine residue between amino acids 24 and 25, compared with the sequence identified for the parents and other siblings and the online reference sequence for a Beagle. Therefore, those authors concluded that this insertion mutation was responsible for the ivermectin-sensitivity phenotype in that Border Collie.10 Interestingly, all 15 of the Border Collies in the present study as well as an Australian Shepherd, a mixed-breed dog, and a Boxer had the reported 3-bp insertion on both chromosomes. Thus, the proposed insertion mutation in the Border Collie of that recent report10 is unlikely to be related to the development of VAM and, in our experience, actually represents the wildtype allele.
Other possible candidate genes for VAM in Border Collies include other drug transport genes or drug metabolism genes such as those for the CYP family. In humans, CYP3A4 and CYP3A5 are involved in the metabolism of vincristine.6 In dogs, the genes responsible for metabolism of vincristine have not been fully evaluated, but it is likely that the CYP3A genes are involved. Two canine CYP3A cDNA sequences (CYP3A12 and CYP3A26) have been isolated and characterized.11 Both CYP3A12 and CYP3A26 would be good candidate genes for the development of VAM in Border Collies.
Dogs with the ABCB1-1Δ mutation, including some Border Collies, are significantly more likely to develop VAM.7 However, ABCB1-wildtype dogs may also develop VAM. In the present study, we found that Border Collies were significantly more likely to develop VAM, compared with the likelihood for dogs of many other breeds. A novel mutation in the coding sequence of ABCB1 is not responsible for the increased likelihood for the development of VAM in Border Collies. Further research is necessary to determine whether pharmacogenetics plays a role in this breed-related adverse drug reaction.
ABBREVIATIONS
| ABCB1 | ATP binding cassette subfamily B1 gene |
| CYP | Cytochrome P450 |
| SNP | Single nucleotide polymorphism |
| VAM | Vincristine-associated myelosuppression |
| VCOG | Veterinary Co-operative Oncology Group |
Veterinary Clinical Pharmacology Laboratory, College of Veterinary Medicine, Washington State University, Pullman, Wash.
Primer3, version 0.4.0, Whitehead Institute for Biomedical Research, Massachusetts Institute of Technology, Cambridge, Mass. Available at: frodo.wi.mit.edu/primer3/. Accessed Mar 1, 2008.
PolyPhen, version 2, Division of Genetics, Brigham and Women's Hospital, Harvard Medical School, Harvard University, Boston, Mass. Available at: genetics.bwh.harvard.edu/pph2/. Accessed Dec 28, 2010.
References
1. Chun R. Lymphoma: which chemotherapy protocol and why? Top Companion Anim Med 2009; 24: 157–162.
2. Rebhun RB, Kent MS, Borrofka SA, et al. CHOP chemotherapy for the treatment of canine multicentric T-cell lymphoma. Vet Comp Oncol 2011; 9: 38–44.
3. Tomiyasu H, Takahashi M, Fujino Y, et al. Gastrointestinal and hematologic adverse events after administration of vincristine, cyclophosphamide and doxorubicin in dogs with lymphoma that underwent a combination multidrug chemotherapy protocol. J Vet Med Sci 2010; 72: 1391–1397.
4. Kaiser CI, Fidel JL, Roos M, et al. Reevaluation of the University of Wisconsin 2-year protocol for treating canine lymphosarcoma. J Am Anim Hosp Assoc 2007; 43: 85–92.
5. Thamm DH, Vail DM. Aftershocks of cancer chemotherapy: managing adverse effects. J Am Anim Hosp Assoc 2007; 43: 1–7.
6. Van Schaik RH. CYP450 pharmacogenetics for personalizing cancer therapy. Drug Resist Update 2008; 11: 77–98.
7. Mealey KL, Fidel J, Gay JM, et al. ABCB1-1D polymorphism can predict hematologic toxicity in dogs treated with vincristine. J Vet Intern Med 2008; 22: 996–1000.
8. Song S, Suzuki H, Kawai R, et al. Effect of PSC 833, a P-glycoprotein modulator, on the disposition of vincristine and digoxin in rats. Drug Metab Dispos 1999; 27: 689–694.
9. Veterinary Co-operative Oncology Group. Common terminology criteria for adverse events (VCOG-CTCAE) following chemotherapy of biological antineoplastic therapy in dogs and cats v1.0. Vet Comp Oncol 2004; 2: 195–213.
10. Han JI, Son HW, Park SC, et al. Novel insertion mutation of a BCB1 gene in an ivermectin-sensitive Border Collie. J Vet Sci 2010; 11: 341–344.
11. Mealey KL, Jabbes M, Spencer E, et al. Differential expression of CYP3A12 and CYP3A26 mRNAs in canine liver and intestine. Xenobiotica 2008; 38: 1305–1312.
Appendix
Exons amplified by primer pairs, sequences of primer pairs, and size of PCR product for amplification of all 27 canine ABCB1 exons.
| Exon | Forward primer sequence (5′ to 3′) | Reverse primer sequence (5′ to 3′) | PCR size (bp) |
|---|---|---|---|
| 1 | GGCAGTGGGGTGAGAACTAGAG | TCAGAGCTGGAGACTAGAAATCC | 453 |
| 2 | GTTTTGTGTAATCCTGTTGCTGTC | TGGAAATCTAGGCCAAGAAGAG | 301 |
| 3 | GGTTGGACCAGGATGGTAATAG | TTTCCCCCAGAAATAAACACAC | 418 |
| 4 | CATTTGAATGGATGCCATAGTG | AAGCCACTTCGAACTCCTCTAA | 463 |
| 5 and 6 | GTTGGCATAGACGTAGTTGGAT | TGGGGACAGTTTAAAACCTAGA | 1,408 |
| 7 | AGATCATCAGAGTCCTGTGTGC | CCCCCACACACATACCTTTATATT | 326 |
| 8 | GCTGGGCACATTCTAAGTCTTT | ATCATCACTCAAGCCAACACC | 346 |
| 9 through 12 | CCATTGTTTCACTGAGCAAGTA | AACAGGCCAGCAATCTATTCTA | 1,298 |
| 13 | TGCTAAGTTTTAAGGACCTGGA | ACTTGACGGGATCACTTTAGAA | 422 |
| 14 and 15 | TGTAAACTTGGGCTGATAGAGGT | GGATAGGACAGGAGGATTTTGA | 1,500 |
| 16 | ATTAAATAGCCAGGGCATCTG | CACAAACTACAAGCACAAAGGA | 696 |
| 17 | TGACTCCCAAATATTGTGCTTG | TTTGGTTCCCAGTAGACCAGTT | 317 |
| 18 | AACATAGCCAACCCTGACATCT | AGGGGTTGTCCAGTCACATATC | 346 |
| 19 | TGAGTGCTAAATGCAGGGATAG | ATAAGGCACATGATTGGTTGTG | 339 |
| 20 | GCTGAATTGTCATCATCCTGAA | AACACCGTTCTCCAAGCATAGT | 403 |
| 21 and 22 | TGGTGAAGAAGCTCATAGGTCA | CGATTGCCAACATTTACAAGAG | 1,693 |
| 23 and 24 | AATTGAGGCAATGCTCTCTAGC | AGGTTGGAAAAACTGTGGCTTA | 1,022 |
| 25 | TCCCATGGTAACCTGACAATTT | GTTTGGAAAACCTGCCTTTTAG | 397 |
| 26 and 27 | CCTGTTTTCTGTGAGCACACTC | ATGATCTTCAGTTGCAGCAAGA | 1,378 |
