Figure 1. Thylacosmilus skull. Note the deep maxillae in dorsal contact containing giant canine roots. These are not present in Patagosmilus.
For those in a hurry: The Janis et al. study includes a phylogenetic analysis of placental sabertooth cats that nests the saber-toothed marsupial, Thylacosmilus (Fig. 1), at the base of the clade (Fig. 2). In a way, that is true, but this is missing so many transitional taxa that we’re left with apples and oranges. So, that’s not going to work because Janis et al. are testing analogy and convergence, rather than homology. It’s better to test apples and apples, even when dealing with stress tests, etc.
Figure 2. Cladogram from Janis et al. 2020. Note the lack of marsupial taxa, other than Thylacosmilus at the base. Be wary whenever the taxon under review nests at the base of the cladogram.
Lacking here is a phylogenetic analysis that includes the closest marsupial relatives of Thylacosmilus: 1. Schowalteria (Fig. 3), 2. Vincelestes + Conorytes, 3. Huerfanodon 4. Monodelphis + Chironectes. That’s how they line up in the large reptile tree (LRT, 1698+ taxa). You need related taxa to decipher the phylogenetic bracketing of Thylacosmilus based on homology, not analogy. The last two taxa are extant. One is an omnivore, the other an aquatic carnivore. Among the extinct taxa, Schowalteria, Vincelestes, Conoryctes and Huefanodon all appear to have been marsupial saber-toothed predators, contra the Janis et al. headline. Vincelestes goes back to the Early Cretaceous.