Answering this needs the eGFR entry criteria of the trial you are quoting, because renal trials enrol by kidney function and the results do not transfer across strata.
The initial eGFR dip is on the order of one to three millilitres per minute per 1.73 square metres and recovers. Reading it as harm and stopping is the error the pattern is designed to catch you with.
Concretely, urinary albumin-to-creatinine ratio falls substantially in this class, often by thirty per cent or more. It is a good surrogate and it is still a surrogate; the reason FLOW mattered is that it measured the thing itself.
Earlier renal findings came as secondary composites within cardiovascular outcome trials and were dominated by the albuminuria component, which is why the dedicated trial mattered.
Albuminuria is the fast signal and the composite is the answer.