Discussion
Despite a rich literature on the relationship between reproduction and lifespan [1], [2], [25], surprisingly little is understood about the mechanisms by which investment in reproduction affects cause of death. Our analysis of causes of death associated with reproductive capability suggests that further and more detailed studies of reproduction and mortality in companion dogs could shed considerable light on this problem. Companion dogs are an established medical model for humans because the two species experience many of the same spontaneously-occurring diseases, participate in analogous high-caliber medical and surgical care, respond similarly to therapy, and share a daily environment [10], [26], [27], [28], [29]. No other species is similarly able to mirror the human experience of the impacts of environment, lifestyle choices, and medical care on health. Furthermore, in North America 50-75% of pet dogs are electively surgically sterilized, as recommended by the American and Canadian Veterinary Medical Associations, the American Society for the Prevention of Cruelty to Animals, and the Humane Society of the United States. The existence of large numbers of reproductively intact and electively sterilized companion dogs provides an unparalleled opportunity to evaluate outcome differences between the groups.
In our study on companion dogs, we identified many underlying causes of death that shape the composite trait of lifespan. In our study overall, lifespan was greater in the sterilized dogs compared with the reproductively intact dogs. While intact reproductive capability was associated with decreased lifespan in dogs, some causes of death were less frequent in intact dogs. Interestingly, we observed the largest-and opposite-effects in two of the most common causes of death among dogs within our dataset: neoplasia and infectious disease. It is not within the scope of our study to determine the causes of these associations. While the absence of gonadal hormones is an obvious physiological outcome of surgical sterilization, downstream consequences of the absence of gonadal hormones, including altered feedback on pituitary or adrenal hormonal axes, or changes in patterns of growth, development, or behavior may also be significant factors.
Sterilization increased the risk of death due to neoplasia, but did not increase risk for all specific kinds of cancer. Female dogs sterilized before sexual maturity are unlikely to develop mammary cancer because of the decrease in cumulative estrogen exposure associated with the absence of the estrus cycle [30]. However, it is not clear why the frequency of some cancers outside the reproductive system, including lymphoma and osteosarcoma, is influenced by sterilization, while the frequency of others, such as melanoma and squamous cell carcinoma, is not. The increased risk of death due to cancer observed in sterilized dogs could be due to the fact that in both sexes, dogs sterilized before the onset of puberty grow taller than their intact counterparts [31] as a result of reduced estrogen signaling [32]. Recent studies in humans suggest that growth is a risk factor for a number of different cancers [33].
Conversely, sterilized dogs had a decreased risk of death due to infection, and avoidance of infection may partly explain their longer lifespans. The relationship between sterilization and infectious disease could arise due to increased levels of progesterone and testosterone [34] in intact dogs, both of which can be immunosuppressive [35], [36]. Studies in humans, mice and rats reveal patterns of infectious disease morbidity and mortality associated with testosterone and estrogen exposure. However, these patterns vary with host species, type of pathogen, and chronicity of infection [37]. Additionally, sterilization and disease risk might both be correlated with specific canine behaviors. Given the opportunity, intact male dogs are more likely than sterilized dogs to roam, and to fight with other dogs, and intact female dogs show more dominance aggression than spayed females [38], [39]. These behaviors might increase the risks of both infectious and traumatic causes of death among intact dogs.
Limited previous studies on the effects of gonadectomy in humans have found some results consistent with ours. Studies in two different populations have shown that eunuchs live longer than their intact male counterparts [40], [41]. Interestingly, Hamilton and Gordon [40] found that the largest factor influencing the survival difference was the high rate of death due to infections among the intact men. However, the study failed to find differences between the two groups in death due to cancer or trauma, both categories in which sterilization was associated with a large effect in our dataset.
Retrospective studies such as this are not without potential weaknesses. For example, elective surgical sterilization and subsequent veterinary care are potentially associated with socioeconomic status. Owners who cannot afford the cost of sterilization might also lack the resources to provide medical care for diseases that later occur, which might result in sterilized dogs who have access to better medical care appearing to live longer. This issue is unlikely to exert a significant impact on our results, however, as all dogs within our dataset were seen at referral institutions, where costs of care are high. Since dogs in our study were owned by people who could afford the cost of referral from their local veterinary practices to specialty hospitals when their pets became ill, it is unlikely that financial resources were a limiting factor in preventive health management choices such as sterilization [42].
A second potential bias is introduced by the fact that our dataset does not provide the age at which each dog was sterilized, the number of times that intact dogs reproduced, or whether sterilized dogs reproduced prior to sterilization. We cannot extrapolate this information from prior work because while most North American veterinary practitioners currently recommend sterilization at 6-9 months of age for pet dogs, and specifically before the first heat cycle in females [11], there is no large study which reports the actual age at which most dogs are sterilized. If the proportion of dogs becoming sterilized were constant within each age bin, then sterilized dogs could appear to live longer simply because the sterilized group steadily expands with increasing age. Previous research has shown that using gonadectomy as a time-dependent variable can give different estimates of the effect of sterilization on longevity than using sterilization status as a straight yes/no response, and that right-censored lifespans from retrospective studies underestimate population lifespan [43], [44]. However, it was not our objective to define the precise life expectancy for any category of dog, merely the difference between two groups that varied only by sterilization status. Both sterilized and intact dogs were subject to the same limitation (i.e., enrollment at the time of death), and the impact on the groups is expected to be proportionate. Furthermore, when causes of death are compared between reproductively intact and sterilized dogs within age bins, the differential effects of sterilization persist. Thus, even in the youngest dogs, when the consequences of sterilization would have manifested over a shorter period of time, an impact is apparent (Figure S1). Regarding parity, we cannot state that all individuals within the sterilized group were nulliparous, nor that all individuals within the intact group had reproduced. Thus, it is likely that there is some crossover between groups with respect to actual reproductive experience. The effect of this crossover, however, would be to minimize any differences identified between groups; thus, the ability to identify a marked difference in lifespan and cause of death risk in the face of imperfect separation of lifetime reproductive experience substantiates the significance of the effect. We also note that parallel patterns of pathophysiologic process morbidity were evident within all age groups, suggesting that these effects are robust to variation arising from differences in parity within and between groups. Nonetheless, future studies would obviously benefit from data on parity in individual dogs.
Finally, as previously mentioned, the link between sterilization and the observed outcomes cannot currently be known. A direct cause-and-effect relationship between reproduction and cause of death is possible, but the actual relationship is likely more complex. In mammals, removal of gonadal hormones has been shown to alter hematological and coagulation parameters, the pituitary-adrenal axis, satiety, neurotransmitters, thymic tissue, and behavior [11], [45], [46], [47], [48], [49], [50]. Any or all of these factors could mediate the differential causes of death observed between the reproductively intact and sterilized dogs of this report. Documentation of these outcome differences now creates the exciting opportunity to investigate the possible causal mechanisms in dogs and other species.
Although a retrospective, epidemiological study such as this cannot prove causality, our results suggest that close scrutiny of specific causes of death, rather than lifespan alone, will greatly improve our understanding of the cumulative impact of reproductive capability on mortality. Our results strongly demonstrate the need to determine the physiologic consequences of sterilization that influence causes of death and lifespan. Shifting the focus from when death occurs to why death occurs could also help to explain contradictory findings from human studies [e.g., 8].