Nature or nurture doesn't matter much once an attack has begun.
I will add this.
From the AVMA site.
Conclusion
Maulings by dogs can cause terrible injuries47 and death—and it is natural for those dealing with the victims to seek to address the immediate causes. However as Duffy et al (2008) wrote of their survey based data: "The substantial within-breed variation…suggests that it is inappropriate to make predictions about a given dog's propensity for aggressive behavior based solely on its breed." While breed is a factor, the impact of other factors relating to the individual animal (such as training method, sex and neutering status), the target (e.g. owner versus stranger), and the context in which the dog is kept (e.g. urban versus rural) prevent breed from having significant predictive value in its own right. Also the nature of a breed has been shown to vary across time, geographically, and according to breed subtypes such as those raised for conformation showing versus field trials.37
Given that breed is a poor sole predictor of aggressiveness and pit bull-type dogs are not implicated in controlled studies it is difficult to support the targeting of this breed as a basis for dog bite prevention. If breeds are to be targeted a cluster of large breeds would be implicated including the German shepherd and shepherd crosses and other breeds that vary by location.
I will add this.
From the AVMA site.
Conclusion
Maulings by dogs can cause terrible injuries47 and death—and it is natural for those dealing with the victims to seek to address the immediate causes. However as Duffy et al (2008) wrote of their survey based data: "The substantial within-breed variation…suggests that it is inappropriate to make predictions about a given dog's propensity for aggressive behavior based solely on its breed." While breed is a factor, the impact of other factors relating to the individual animal (such as training method, sex and neutering status), the target (e.g. owner versus stranger), and the context in which the dog is kept (e.g. urban versus rural) prevent breed from having significant predictive value in its own right. Also the nature of a breed has been shown to vary across time, geographically, and according to breed subtypes such as those raised for conformation showing versus field trials.37
Given that breed is a poor sole predictor of aggressiveness and pit bull-type dogs are not implicated in controlled studies it is difficult to support the targeting of this breed as a basis for dog bite prevention. If breeds are to be targeted a cluster of large breeds would be implicated including the German shepherd and shepherd crosses and other breeds that vary by location.