The Marek scenario occurs when a vaccine removes the evolutionary pressure against more virulent strains.
The actual mutation event that creates the more virulent strain occurs randomly. Whether that variant survives and thrives depends on all of the evolutionary pressures in force. If a strain is so virulent that it kills the host, it is selected against, because the dead can't spread. If a vaccine lets the host survive but does nothing to stop transmission, that negative evolutionary pressure is gone. Thus, the potential for a vaccine to drive the propagation of a more virulent strain if all other things are equal.
However, there's good reason to believe that all other things will rarely be equal.
The survival rate of the host is just one selection pressure. Anything can be a selection pressure if it has an impact on reproductive success of a subset of the population. It's possible that increased host survival is a very tiny or even zero pressure - for example, if the virus has a long contagious period, it might have already done all the spreading it's going to do before onset of symptoms and death, rendering survival time irrelevant. Other pressures could dominate - for example, the virus could be susceptible to being killed by sunlight, favoring variants with a light-insensitivity property over variants with a virulence property.
One could even imagine how host symptoms could create positive pressure towards virulence - for example, more coughing and bleeding could help transmission. Such a virus would have its survival chances reduced by a leaky-but-symptom-reducing vaccine, and increased by vaccine-avoidance.
Evolution is spectacularly multi-dimensional. None of these outcomes are inevitable.
The actual mutation event that creates the more virulent strain occurs randomly. Whether that variant survives and thrives depends on all of the evolutionary pressures in force. If a strain is so virulent that it kills the host, it is selected against, because the dead can't spread. If a vaccine lets the host survive but does nothing to stop transmission, that negative evolutionary pressure is gone. Thus, the potential for a vaccine to drive the propagation of a more virulent strain if all other things are equal.
However, there's good reason to believe that all other things will rarely be equal.
The survival rate of the host is just one selection pressure. Anything can be a selection pressure if it has an impact on reproductive success of a subset of the population. It's possible that increased host survival is a very tiny or even zero pressure - for example, if the virus has a long contagious period, it might have already done all the spreading it's going to do before onset of symptoms and death, rendering survival time irrelevant. Other pressures could dominate - for example, the virus could be susceptible to being killed by sunlight, favoring variants with a light-insensitivity property over variants with a virulence property.
One could even imagine how host symptoms could create positive pressure towards virulence - for example, more coughing and bleeding could help transmission. Such a virus would have its survival chances reduced by a leaky-but-symptom-reducing vaccine, and increased by vaccine-avoidance.
Evolution is spectacularly multi-dimensional. None of these outcomes are inevitable.