Materials and Methods

Animals—All 27 cats with VAFS included in this study were domestic shorthair cats referred to the University of Minnesota Veterinary Teaching Hospital for surgical treatment of primary tumors. All tumors were located at sites of prior vaccination and confirmed by board-certified pathologists to have presented typical histopathologic features of vaccine-associated sarcoma.^6,10,19 Tumors were located in the interscapular or scapular area for 19 patients, whereas 8 were located on the proximal portion of a hind limb (upper portion of thigh). All of the 27 patients were undergoing their first treatment for VAFS, and thoracic radiographs available at the outset indicated that the patients were free of lung metastases at the time. Wide (3 to 5 cm) margins were maintained during surgical resection of the tumors and described as radical in 8 cats (removal of dorsal spinous processes in 6 cats and amputation in 2 cats).¹⁵ None received adjuvant chemotherapy or radiation as part of treatment for the primary tumor. Demographics, vaccination history, and clinical details were recorded for the study.

Tissue collection and storage—Specimens of the tumors and tumor-adjacent nonmalignant tissue were taken from incidental materials generated during surgical resection of the tumors by board-certified veterinary surgeons. The tissue specimens were snap-frozen and stored at −80°C until processed for DNA extraction. Blood samples collected from the cats in evacuated tubes^a were stored overnight (if necessary) at 4°C prior to DNA extraction.

DNA extraction—Genomic DNA was extracted from the samples by use of a commercially available kit for DNA purification.^b Genomic DNA from the blood of a disease-free cat was purchased^c and used as a positive control for PCR analysis.

Amplification of p53 exons 5 to 8 by PCR assay— Segments of feline p53 were amplified by use of 3 pairs of site-specific primers designed from the primary genomic sequence (Appendix; GenBank accession Nos., D26608, DQ119105, and AF175762).^20,21 Each reaction mix contained 50 ng of genomic DNA template, 0.2μM each of forward and reverse primers, 100μM deoxynucleotide triphosphates, 1.5mM MgCl₂, 0.75 units of Taq,^d and 1X PCR buffer^e in a total volume of 25 μL. After activation of Taq^d at 94°C for 10 minutes, amplification was performed by use of the first 2 primer pairs with annealing temperatures of 65°C for 2 cycles, 60°C for 3 cycles, and 55°C for 25 cycles for the first 2 primer pairs. An annealing temperature of 55°C was maintained for 30 cycles for amplification by use of the third primer pair. Extension and denaturation temperatures were maintained at 72°C and 94°C, respectively, and all segments were 30 seconds in duration except on the last cycle, for which the extension period was increased to 5 minutes.

Nucleotide sequence analysis—Amplicons were purified by ultrafiltration^e and sequenced^f,g for all study cats. Sequencing reaction mixtures were electrophoresed on an automated DNA sequencer^h at the University of Minnesota Advanced Genetic Analysis Center and sequences analyzed by use of software programs.^i,j

Follow-up examinations—Patients were generally examined at 2 weeks and 1, 3, 6, 9, 12, 18, 24, 30, and 36 months after surgery. Owners also monitored the surgical site for lumps or other abnormalities and brought their cats to the clinic for physical examination in case of concerns in between scheduled visits. Biopsies were performed on recurrent masses for detection of malignant cells. Thoracic radiographs were performed 3 months after surgery and repeated at follow-up visits if metastatic disease was suspected. All cats were followed for at least 3 years after surgery, and cats that died within the period were subjected to necroscopic examination.

Data analysis—Disease-free intervals were calculated as the days from date of first surgery to local or distant disease recurrence (if any). Overall survival was also calculated from date of first surgery to death if the event occurred. Mean, median, and SD values were determined for demographics (such as age and weight) and tumor size. Event distribution was plotted by the Kaplan-Meier method. Data were censored for 8 cats that were in good health with no occurrence of event as well as 1 cat that died of other causes during the 3 years of follow-up. No cats were otherwise lost to follow-up. Differences in DFI for the cats that harbored a p53 deletion (LOH-positive) and cats without the somatic loss in their tumor tissues (LOH-negative) were compared by use of a logrank test (a nonparametric test suitable for the number of samples included in the study).

The associations between DFI and LOH, age, sex, and weight of the cats as well as tumor size, tumor site, or the time elapsed from vaccination to tumor development at the site were assessed with a Cox proportional hazards regression model with regression coefficients determined for each covariate and risk ratios calculated from the exponent of the coefficients. Analyses were performed by use of statistical software packages.^k,l Values of P < 0.05 were considered significant.

Results

Animals—The study included 11 spayed female and 16 neutered male domestic shorthair cats. All 27 cats had a diagnosis of VAFS on the basis of previously described histologic features and resembled cats in larger epidemiologic investigations in their demographic information.^5,6,10 The mean age of the study population was 8.7 ± 3.1 years (range, 1.5 to 16 years). The median age was 8 years, with all but 2 cats within the range of 4 to 13 years. The mean weight of the cats was 5.6 ± 1.3 kg, (median, 5.6 kg; range, 3 to 8 kg). Cats with VAFS had a history of vaccine administration against rabies, FeLV, feline viral rhinotracheitis, calicivirus, and panleukopenia. The vaccine-site tumors developed in 55% of the cats within 1 year of vaccination, and 92% of cats had tumors by the third year. The median period between vaccine administration and tumor development at the site was 12 months (range, 1 to 72 months). On average, tumors were 4.3 ± 2.9 cm at their largest dimension (median, 3.3 cm; range, 1.5 to 19.0 cm). Margins after resection of primary tumors contained malignant cells (ie, not clear margins) for 9 cats. No malignant cells could be detected in the margins (ie, clear margins) for 10 cats; margin status was unknown for 8 cats.

Polymorphisms in feline p53 sequence— Automated DNA sequence analysis of p53 amplicons from samples obtained from the study cats and the control feline genomic DNA revealed the presence of 7 SNPs in addition to a previously reported polymorphism at the third base of codon 163 in exon 5 (GenBank accession Nos., DQ119105 and AF175762).²² Of these sites, polymorphism was prevalent for the third base of codon 163 (C and T); intron 7 positions 14 (C and T) and 259 (C and T); and intron 8 positions 15 (C and T), 20 (A and G), and 70 (C and T). Two sites, intron 7 position 12 and intron 8 position 16, mainly had alleles with C and G nucleotides (instead of T and A), respectively. In addition, insertion of a third thymidine to the existing 2 thymidines at positions 246 and 247 of intron 7 was also observed. Of the 27 cats, 7 were homozygous at all the polymorphic sites and thus were noninformative for LOH in the conserved domain of p53. The remaining 20 cats were classified as informative (ie, heterozygous for at least 1 polymorphic site and therefore informative for allelic loss at the locus in malignant cells).

Somatic alterations of p53in VAFS—Analysis of the polymorphic sites in DNA from blood samples and tumor specimens from the 20 informative group cats revealed allelic loss of p53 in 12 tumors. In addition to LOH, point mutations in exon 5 were detected in the primary tumors from 3 cats. Of these, 1 tumor had a missense mutation at codon 179, which is located within a region of p53 that is conserved across species.²³ The wild-type sequence at codon 179 encodes a histidine residue that plays an important role in maintaining the 3-dimensional structure of the core domain of p53.²³ A transversion at the first position of the codon (from C to A) with a predicted amino acid sequence change to asparagine occurred in the single retained allele in this tumor. Another tumor had a single base deletion at the third position of codon 184, leading to a series of amino acid substitutions within the evolutionarily conserved domain (from codon 184 onward) and culminating in a stop codon in exon 7. The third mutation consisted of a transition (from G to C) at the second base of codon 156 (a location that is not well conserved across species), leading to an amino acid change from cystine to serine.

Follow-up data—By 3 years (1,095 days) of follow-up, 18 of the 27 cats (66.7%) had died or been euthanatized with disease, 1 had died of other causes at 111 days after surgery, and 8 (29.6%) were alive and in good health with no disease recurrence. The 18 cats with disease relapse included 2 cats that had metastasis (145 and 677 days after surgery, respectively), 3 cats that developed local recurrence as well as metastases, and 13 that had local disease only. Thus, the overall rates of metastasis and local recurrence for cats in this study were 18.5% and 59.3%, respectively. All 5 cats with metastasis were euthanatized at the time of detection of lung involvement, whereas 8 of the 13 cats with only locally recurrent disease received further surgical treatment (along with adjuvant treatment with γ radiation in 3 cats and intralesional injection of carboplatin in 1 cat).

Impact of allelic deletion of p53on outcome—

Allelic deletion in p53 in the primary tumor had a negative impact on outcome as shown by the Kaplan- Meier plot (Figure 1). Cats that were LOH positive in the primary tumor tissue rapidly developed recurrences (median DFI, 196 days; range of DFI, 39 to 311 days). Of these 12 cats with LOH-positive primary tumors and disease recurrence, 3 had developed local and distant disease at the time of recurrence, whereas 1 developed distant metastases without signs of local recurrence. Log-rank analyses indicated significant differences in DFI between the LOH-positive group and cats that were either LOH negative (P < 0.001) or non- informative for LOH (P = 0.010). Five of 8 cats in the LOH-negative group had no relapse after surgery at > 3 years of follow-up, 2 developed local recurrences at 315 and 292 days after surgery, and 1 cat developed metastasis detected at 677 days after surgery. Although all the primary tumors in this group were LOH negative, the tumor that recurred at 292 days (the shortest DFI in this group) contained the mutation at codon 156 and later underwent allelic loss of p53 in the recurrent tumor. In the noninformative group, 3 of 7 cats had no relapse after surgery at > 3 years of follow-up, 1 died of other causes at 111 days into the follow-up period, and 3 developed local recurrences (at 46, 105, and 492 days, respectively).

View Full Size

Kaplan-Meier distribution curves comparing outcomes on the basis of p53status of primary tumors. Cats with allelic deletion of p53in their primary tumors are represented as LOH positive, whereas cats with no allelic deletion are grouped as LOH negative. Cats for which LOH could not be evaluated are presented as a separate group marked as noninformative for LOH. The DFI was calculated as the days from date of first surgery to local or distant disease recurrence (if any). Differences in DFI between the LOH-positive group cats and LOH-negative and noninformative group cats were significant (P < 0.001 and P= 0.010, respectively). Hatch marks represent censored data for 8 cats (5 in the LOH-negative group and 3 in the noninformative group) that were in good health with no occurrence of event by the end of the 3-year follow-up period, as well as 1 cat (in the noninformative group) that died of other causes during the 3-years of follow-up. LOH – = LOH negative. LOH + = LOH positive. Ni = Noninformative.

Citation: American Journal of Veterinary Research 67, 10; 10.2460/ajvr.67.10.1766

• Download Figure
• Download figure as PowerPoint slide

As with DFI, a significant negative association was also found between occurrence of LOH in the primary tumor and overall duration of survival (Kaplan-Meier plot not shown). At 3 years of follow-up, 14 of 20 informative group cats (12 of 12 cats from the LOH- positive group and 2 of 8 cats from the LOH-negative group) had died or had been euthanatized with disease. Cats in the LOH-positive group had an overall median survival of 223 days (range, 51 to 730 days). Survival was significantly (log-rank test, P < 0.001) longer for cats that did not harbor LOH in their primary tumors, with 6 of the 8 cats in this group alive at > 3 years. Two cats in this group had been euthanatized, 1 at 677 days because of development of metastasis and 1 at 748 days because of aggressive local recurrence of the tumor after a second surgery with adjuvant radiation treatment. The primary tumor from this latter cat contained p53 mutation at codon 156, and the first recurrence was diagnosed at 292 days after the initial surgery. Survival was also significantly (log-rank test, P = 0.030) longer in the noninformative group, compared with the LOH-positive group. In the noninformative group, 3 of 7 cats were alive with no disease at 3 years past initial surgery, 1 died of other causes before the end of the follow-up period, and 3 with locally recurrent disease were euthanatized (at 110, 492, and 546 days after initial surgery, respectively).

The significant differences in overall survival between the LOH-positive group and the LOH-negative or noninformative groups were mainly attributable to the presence of cats with no disease recurrence in the latter 2 groups and could be detected despite the additional treatment provided to the 8 cats with recurrent disease in the study (which may have affected survival in those cats). Although the number of cats with recurrent disease receiving additional treatment was not enough to allow evaluation of the different treatment modalities, successive treatment for recurrent disease produced a significantly (log-rank test, P = 0.034) better outcome for the 2 LOH-negative cats than the 6 that belonged to the LOH-positive and non- informative groups. One cat receiving treatment for recurrent disease in the LOH-negative group achieved 414 disease-free days before the second relapse, whereas another cat in this group was healthy and disease free at the end of the 3-year period. In contrast, successive treatment resulted in much shorter times to second recurrence in LOH-positive and noninformative groups (median, 52 days; range, 24 to 122 days).

Impact of other factors on disease outcome–—

Although cats in the LOH-positive group had larger median values for age, weight, tumor size, and time elapsed from vaccination, regression analysis by the Cox proportional hazards model indicated no significant associations between these parameters or sex and disease relapse (age, RR = 1.03, 95% CI of 0.83 to 1.28, P = 0.767; weight, RR = 1.18, 95% CI of 0.8 to 1.75, P= 0.387; tumor size, RR = 1.32, 95% CI of 1.06 to 1.66, P = 0.078; time from vaccination, RR = 1.02, 95% CI of 0. 34 to 1.29, P = 0.842; sex, RR = 2.36, 95% CI of 0.49 to 11.36, P = 0.290). Adding whether the surgical procedure was described as radical or not and tumor location (described as scapular area or proximal portion of hind limb) to the analysis covariate did not provide indication of any significant association. However, the strong association between LOH in the primary tumor and disease relapse was confirmed among the group of covariates (RR = 42.35, 95% CI of 4.38 to 409.5, P = 0.001). No significant relationship was detected between LOH status and margin information. Although a significant (Cox proportional hazards regression model, P = 0.022) association with shorter DFI could be detected when malignant cells were present in the margins, DFI bore no significant association with margins when they were thought to be devoid of malignant cells. Thus, although disease relapse could be predicted correctly in 8 of 9 cats by margin information when margins contained malignant cells, 4 relapses occurred in the 10 cats that were thought to have clear margins. Because tumor size may have impacted the ability to obtain margins devoid of malignant cells, we further examined the subset of 13 cats for which the tumors were ≤ 3 cm (median, 2 cm; range, 1.5 to 3.0 cm) in their largest dimension. Information on margins could be retrieved for 9 of these cats but did not correspond with outcome in 4 instances. Three cats with apparently clear margins had tumor recurrence; 1 cat in which malignant cells were thought to be in the margin had no tumor recurrence. p53 status could be determined in the primary tumors for 10 cats and correctly predicted disease outcome in each instance.

Discussion

To begin to elucidate the molecular pathogenesis of VAFS, we have analyzed primary tumors in cats with VAFS for somatic alterations in p53. The evolutionarily conserved domain of this tumor suppressor gene is the most frequent site of somatic mutations in various cancers including sarcomas of the soft tissue.^23–26 Analysis of the genomic sequence of feline p53 revealed the presence of 8 polymorphic sites. These included a previously identified SNP at codon 163 of exon 5, 3 sites in intron 7, and 4 sites in intron 8. Such intronic polymorphisms of p53 are also present in other species, and while the functional importance of individual bases at each of the sites is not known, the presence of distinct bases at these sites (on each of the 2 copies of the gene normally present in an individual) helps to distinguish the alleles from each other.²⁴ Hence, comparison of sequence information at polymorphic sites in DNA obtained from matched sets of malignant and nonmalignant tissues from individual patients enables detection of allelic deletions in the malignant cells through measurement of LOH at each locus. Because individuals may be homozygous (ie, noninformative for LOH) at some of the polymorphic sites, analysis of multiple sites increases the chances that > 1 site will be informative.

Results of our study on polymorphic sites in primary tumors of cats with VAFS provide strong evidence for the involvement of allelic deletion at p53 in the malignant progression of VAFS. Sixty percent (12/20) of the informative group cats had undergone the allelic deletion (as evidenced by LOH) in their primary tumors. The presence of LOH was found to be associated with a significantly (log-rank test, P < 0.001) increased frequency of disease recurrence after aggressive surgical treatment as well as significantly (log-rank test, P < 0.001) shorter overall survival. In the LOH-positive group, 6 of 12 cats relapsed by 4 months, 9 of 12 cats did so by 6 months, and all 12 cats had relapsed by 10 months. In contrast, only 3 of 8 cats had recurrence in the LOH-negative group: 2 had local recurrences at about 10 months after surgery, and the third had distant metastasis at > 22 months.

Significant differences in DFI (log-rank test, P = 0.010) and survival (log-rank test, P = 0.030) were also found between the LOH-positive group cats and non- informative group cats. It is possible that the 3 cats from the noninformative group that had recurrent disease also harbored p53 deletions in their primary tumors but that these alterations could not be detected through examination of the SNP sites located within the p53 genomic sequence between exons 5 through 9. This possibility is supported by our observation that the DFIs were similar after a second treatment for recurrent disease for cats from noninformative and LOH-positive groups. Expansion of the feline p53 SNP repertoire in the future could help in reducing the number of cats that are noninformative for LOH.

Missense point mutations were detected in only 3 of the 27 primary tumors examined by sequence analysis of exons 5 through 9 of the p53 tumor suppressor gene. All 3 of these cats serve as classic example of Knudson's theory of inactivation of both copies of a tumor suppressor gene in malignant cells.²⁷ In 2 instances, alterations occurred on the single p53 allele retained in aggressively growing tumors that had undergone LOH. In the third instance, the point mutation appeared first (in the primary tumor) and was followed by deletion of the other allele in the corresponding recurrent tumor. The presence of somatic point mutations of p53 has also been reported by other investigators in 5 of 8 tumors chosen from a larger set of 40 unclassified vaccine-site tumors on the basis of nuclear immunohistochemical staining for p53.²⁸

Overall, results of our study indicate that in VAFS, the first allelic inactivation of p53 is likely to occur through deletion and that this may be followed by inactivation of the remaining allele through accumulation of point mutations. Occasional point mutations of p53 are also capable of leading to a more aggressive tumor phenotype, but this is likely dependent on the specific location of the point mutation and the type of change it brings about in the translated protein. Although it is possible that other mechanisms, such as amplification or enhanced translation of Mdm2, or gross rearrangements outside the conserved regions are also involved in the inactivation of p53, loss of 1 allele appears to be a crucial switch (or at least marker) associated with acquisition of invasive characteristics and malignant progression of VAFS.^26,29 It has been our observation that malignant cells with reduced or altered tumor suppressor activity can invade deep into the surrounding tissue; even a minimal number of such cells remaining after surgery may multiply in the environment of inflammation and wound healing at the site and rapidly grow into the recurrent tumor.^30–32

Results of our study further indicate that although the presence of malignant cells in the margins after surgical resection of primary tumors is a significant (Cox proportional hazards regression model, P = 0.023) predictor of negative outcome, tumors resected with apparently clear margins can recur as well and that such recurrence may be predicted by a knowledge of p53 status of the tumor. This lack of significant association between completeness of excision and outcome has also been found in at least 2 other larger studies (based on 78³³ and 92³⁴ cats with VAFS, respectively). The study³³ on 78 cats with VAFS revealed that recurrence rate or survival was not associated with completeness of excision regardless of whether the margin information was based on the surgeon's impression or results of histologic evaluation. Also, in the study³⁴ on 92 cats with VAFS, local recurrences were found to develop in 42% of cats with complete excision of tumors (ie, with clear margins). Results of our study further indicate that knowledge of p53 status is important even for the management of relatively small tumors. For the 13 cats included in our study in which the primary tumors were ≤ 3 cm (median, 2 cm) in their largest dimension, p53 status was a far more accurate predictor of disease outcome than margin information. It is possible that the initiation of malignant progression to an invasive phenotype through allelic deletion of p53 may occur relatively early and before the appearance of severe morphologic changes in the tumor cells.

Finally, results of our study indicate that the analysis of p53 status in primary tumors provides critical information for the management of VAFS irrespective of tumor site. Unlike p53 status, the tumor site per se (scapular or proximal portion of the hind limb) did not significantly affect outcome in our study, wherein all tumors were removed with an aggressive, well-planned surgery at first chance. In a previous study¹⁵ aimed at identifying populations of cats that might be effectively treated with surgery alone, it had appeared as if tumors on the limbs were associated with a significantly longer time to first recurrence (median, 325 days) than tumors of the trunk (median, 66 days); however, as clearly pointed out by the authors of that study, such comparisons were not possible because the tumors located on the limbs underwent radical excision (amputation), whereas the tumors that were located on the trunk were mostly removed with marginal first excisions performed prior to referral of the patients to an advanced veterinary center.

During the last 15 years, vaccination has been etiologically linked with the development of aggressive sarcomas in cats. Although the disease develops in 1 to 13 instances for every 10,000 vaccines administered, the poor prognosis of the disease has caused great concern to veterinary health providers and pet owners. Unfortunately, and despite the great interest in the field, not much information has been available on the molecular pathogenesis of the disease.³⁵ Aggressive surgery at first chance at a well-equipped hospital is known to be the most effective mode of treatment, whereas adjuvant treatment with γ radiation as well as various chemotherapeutic agents is also often used in efforts to better manage the disease.^15,35–38 However, the efficacy of continued treatment or the need for such treatment among different groups of cats with VAFS is not known.¹⁵ Results of our study indicate that in the subset of cats without allelic loss of p53 in their tumors, aggressive surgical treatment could provide a positive prognosis in absence of additional treatment. Our results further indicate that cats with p53 alterations in the primary tumor are likely to have recurrences within a year despite aggressive surgical resection. The prognosis for such affected cats is likely worse when the p53 alteration consists of allelic deletion, a relatively common event in the pathogenesis of VAFS. Although the efficacy and benefit of various modalities of continued treatment of recurrent disease are not yet known for VAFS, evaluation of p53 status in primary tumors may be a useful tool in screening for cats that are likely to have any added benefit from additional intervention.

As genes that play a role in the molecular pathogenesis of VAFS get elucidated (as is beginning to happen for various human sarcoma subtypes), additional markers and molecular targets could be developed for individualized treatment and improved management of the disease.^39,40 In the interim, early detection of the disease, prior to the onset of p53 alterations, through vigilant monitoring of injection sites likely provides the best means to a positive outcome. It is likely beneficial to also continue the practice of not vaccinating in the scapular region because of the practical limitations of performing surgical excision of tumors with wide margins at the site. Furthermore, efforts and resources are necessary to perform large well-planned studies wherein the large number of clinical parameters that affect outcome (such as specifics of treatment subgroups, use of comprehensive diagnostic tools, and follow- up processes used) can be carefully controlled and analyzed. Above all, research on the subject and publication of results should be encouraged so as to help alleviate this important companion animal health issue.^19,35