I have to discuss the question, why anyone should still use 2D cell culture, as this tumor model seems to be the worst of all models. Do you see any advantages?
There "seems to be" are two "advantages" for 2D models, which in fact are not at all advantages. Low costs and easy to use. But totally unrepresentative of what happens in a cancer in a patient.
It is not because "many people are doing something" that this thing is "good".
And most of the people using 2D models while claiming that they obtained "prominsing anticancer effects" should question themselves about the absence of actually active new (since 20 years) compounds against metastatic cancers, which still kill 90% of the cancer patients.
3D, GEMMs, PDX are more complex. However, are they closer to reality? And the other question: are 2D really obsolete? Or do they still play a role in e.g. proteomics or genomics?
I (unfortunately) reached a "significantly" older age than you ... thus I have time to read hundreds of articles and sometimes I am lucky enough to publish reviews ...
First of all, a model remains a model.
Second, there is no "wrong" versus "right" or "good" versus "bad" models. A model is the only way to scientifically validate an hypothesis.
Thus, for sure 1) 2D models can be good models for certain questions in proteomics or genomics, while they can be bad models for questions in the same areas. Once more, it is the researcher who will make a good versus a bad choice to scientifically demonstrate her / his hypothesis.
I am still often using (when I am not writing reviews ..) 2D models.
A fact in which I am nevertheless convinced at the scientific level is that cancer cells associated with cancer-associated cells (strange sentence ...) and developing in a 3D environment will have a biological behavior closer to the "biological reality" than cancer cells growing alone in a 2D environment.
So, I am convinced that mutated genes ar conserved in 2D cancer models, but I am also convinced that the sigaling pathways directly or even undirectly under the control of these mutated genes will not be the same in 2D and 3D models.
In the same manner, I am convinced that intracellular signal processings will (very often, in not in most case) markdly differ between the same cancer cell line developing alone in a 2D model versus developing in a 3D model with cancer-associated cells (not only the fibroblasts but also the amazing importance of tumor immune-related cells), because of the influences (only few examples among so many ones) of the actin cytoskeleton, of the ECM elements, of the way of migration, etc., etc...
As a conclusion, all these remarks include the fact that i) my comments result of my own and only own interpretation about what I red, 2) a model is a model and there is no "bad" versus "good" model, 3) and now an approximate English translation of what we say in French: "we do not need a big gun to kill a mosquito", and 4) may be the most important, no one is the owner of the "scientific truth", certainly in the cancer area.