This topic does not explain the fatigue limit that leads to pitting. It is limited to the lower end of the FatigueLife, approximately ≤103 cycles.
As we increase the contactforce/payload between the cam and the camfollower, the weaker material will eventually deform plastically rather than elastically.
To specify the critical load at which the weaker material 'plastically fails', we need to use a yield criterion. We can find the yield of metals with the 'Tresca Maximum Shear Stress' or the 'vonMises Strain Energy' failure criteria. We will use the Tresca Maximum Shear Stress Failure Criterion.
Simplified, the Tresca Shear Stress failure criterion tells us that Yield occurs due to Shear Stress when the Shear Stress becomes equal to half the Yield Strength that results from a normal tension test.
$k=\frac{Y}{2}$ Y = Yield Strength k = Shear Strength Yield strength usually has the label as Re in material tables that yield, or Rp0.2 for brittle materials that do not have a yield point. 
We can use the Contact Stress with the Maximum Shear Stress failure criterion to determine the onset of plasticity. We know that ... ${\tau}_{\mathrm{max}}=0.3\times {P}_{\mathrm{max}}$ Thus, when ... ${\tau}_{\mathrm{max}}=k$ ... then ... ${\tau}_{\mathrm{max}}=k=0.3\times {P}_{\mathrm{max}}=\frac{Y}{2}$ Then, we state that the Onset of Plasticity, for linecontact, occurs when... $k\u22150.3={P}_{\mathrm{max}}=Y\u22152\times 0.3$ or, equivalently ... ${P}_{\mathrm{max}}=3.3k=1.67Y$ 

Summary LineContact Plasticity starts, as a tiny pinpoint, when:
Plasticity starts at a depth of
For Elliptical Contact, the Stress at which plasticity starts is not very different to that with LineContact. However, the depth at which Plasticity starts reduces to ~0.33b. 
When = 1.67 × Y [τmax = Y/2] there is an 'onset of plasticity' [yield begins] that starts at 0.78b below the surface. Even if some yield takes place, there will only be a small change to the shape. This is because the yield occurs beneath the surface and the plastic zone is surrounded by a region in which the stress and strains remain elastic. This limits the plastic deformation since the plastic strains must be similar to adjacent elastic strains. The amount of steel that is flattened [moved] by Plastic Strain under the contact must be equal to the expansion of the surrounding steel with Elastic Strain. The first time that the elastic limit is exceeded at a point on the cam, a small plastic deformation occurs. When the camfollower is further along the cam, and the load becomes zero again, residual stresses remain in the cam, below the surface. When the camfollower passes the same point on the cam a second time, the cam stress is the combined action of contactstress + residual stress from the first pass. Because residual stresses are usually 'protective', the cam is less likely to yield when the camfollower makes a second pass than the cam was when the camfollower made its first pass. [ Note: according to theory not presented here: 'if it is possible for residual stresses to occur, they will always occur']. It is possible, that after a few rotations of the cam, the contact becomes elastic, because of the buildup of the protective residualstresses. This process is called Elastic Shakedown. The new ElasticLimit after 'shakedown' can increase to: 4.0 × k , compared to 3.33 × k for the ElasticLimit before ElasticShakedown. Other factors that help Shakedown: StrainHardening: where the cam strains, it may 'StrainHarden'. Thus, its yieldstrength will also increase. Note: StrainHardening is not ElasticShakedown. Steels that already very hard may StrainWeaken. GeometricChanges: if the camsteel deforms, then the cam and camfollower become more 'conformal'. A flat cam becomes cupshaped when the camfollower is barrelshaped, for example. The conformity decreases the contactstress because the Equivalent RadiusofCurvature, Re, increases. It turns out that the Hertzian Contact Stress, must be less than or equal to 4.00×k for Elastic Shakedown to occur. This means that if the contact stress is: ...then, Elastic Shakedown will occur. Also, as the load F is proportional to , it follows that the ratio of the Shakedown Force limit, , to the Elastic Force Limit, , is given by This means that the actual load can be increased by ~50% more than the load that is needed to bring about yield on the first machine cycle. 
As the contact stress increases, the plastic zone grows until eventually it reaches the cam surface. Full Plasticity occurs when the Contact Stress reaches the surface. The Contact Stress is then: The cam will fail with Low Cycle Fatigue, ~103 cycles. The theory that predicts this is 'upper bound' theory, which postulates plastic flow. It equates the rate at which the external load does work to the rate at which energy is absorbed within the material  defined as the upper bound of the true maximum load the contact can sustain. [Personally, I have no idea about this theory]. 
A : Upper Bound to elasticshakedown limit against alternating plasticity B : Upper Bound to plastic shakedown limit against incremental growth. C : Upper bound to elastic shakedown limit against incremental growth of surface strain E : Elastic limit  Lower bound to elastic limit. 

Traction Coefficient, μ, has an effect on the maximum value and the depth of the maximum shear stress. If the camfollower both slides and rolls over the cam, then the friction force moves the maximum shear stress closer to the surface as well as increasing its value. When traction or friction is zero, the maximum shearstress is at 0.78×b [contact half width]. At a CoefficientofFriction μ = 0.3, the maximum shear stress moves to, or very near to, the surface. Thus, plasticity is not constrained by the elastic hinterland. You can see in the image, that the range between the stress when there is 'first yield' (Blue line) and that for 'shakedown' (Magenta line) becomes narrower as you increase the coefficient of friction and there is sliding at the contact. 