As usual there is a vast range of contradictory outcomes when looking at drug uses of any type. I have seen various statements that its protective for cancer, others that it increases dementia . So basically it might be very good, very bad, or somewhere in between depending on what the user is actually vulnerable to depending on their own genetic make up. .
As in all things, if the drug in question does something that you can do via natural food instead, I would opt for the natural food as my first resort and then move onto the drugs if I had to. Which is why think the advice should be to try through diet first so you can see exactly what difference the metformin actually makes rather than simply prescribing it as a first option.
Here it is represented a a wonder drug - reducing cancer risk
https://www.diabetesselfmanagement.com/blog/metformin-wonder-drug/
primarily through bringing down blood sugar and insulin - both often acheivable through diet
Here is is represented as a very bad drug which increases the risk of alzheimers and dementia
https://www.diabetesdaily.com/forum/diabetes-news-and-studies/96797-metformin-very-bad-if-true/
Table 1. Outcomes in Metformin Users vs Nonusers
Outcome Event Rate (%) Adjusted Hazard Ratio (95% Confidence Interval)
PD 6.85 vs 2.78 2.27 (1.66 - 3.07)
All-cause dementia 11.5 vs 6.7 1.66 (1.35 - 2.04)
Alzheimer's dementia 1.64 vs 0.83 2.13 (1.20 - 3.79)
Vascular dementia 1.64 vs 0.69 2.30 (1.25 - 4.22)
In addition, the outcomes mostly increased as the use duration and dosage of metformin increased, especially with use for more than 300 days and doses greater than 240 g.
Table 2. Comparison of Metformin Use Duration vs Nonuse
Risk Event Rate (%) Adjusted Hazard Ratio (95% Confidence Interval)
For PD
<180 d 5.90 1.77 (1.17 - 2.68)
180 - 300 d 4.30 1.46 (0.90 - 2.37)
300 - 400 d 6.05 2.20 (1.47 - 3.28)
≥400 d 14.3 4.49 (3.06 - 6.58)
For all-cause dementia
<180 d 7.99 1.02 (0.74 - 1.41)
180 - 300 d 11.4 1.79 (1.32 - 2.43)
300 - 400 d 10.4 1.61 (1.21 - 2.16)
≥400 d 20.6 2.84 (2.12 - 3.82)