An 18-month multimodal intervention combining exercise, nutrition, and metabolic management produced significant cognitive preservation — but only in a specific subgroup. Among 188 Japanese older adults in the J-MINT PRIME Kanagawa trial, a CART-identified 'Target Group' (N=108) characterized by baseline MMSE scores below 28 or LDL-C below 135 mg/dL showed meaningful MMSE trajectory preservation versus controls (group × time interaction, P=0.022). The 80 high-functioning participants (MMSE ≥28) showed no benefit.
This finding slots into a well-established pattern across dementia prevention trials: people closest to the cognitive cliff benefit most from intervention, while those with high baseline function have little room to show measurable improvement — the so-called ceiling effect. The FINGER trial and subsequent multimodal programs have faced similar interpretive challenges. What's notable here is the use of CART analysis to prospectively define a responsive phenotype incorporating both cognitive status and metabolic markers, suggesting that LDL-C level may signal vascular vulnerability amenable to lifestyle correction.
Several cautions apply. The primary trial showed no overall benefit, making this a post-hoc, hypothesis-generating analysis prone to overfitting. The cohort is modest at 188 participants, and the Japanese population limits generalizability. Practice effects on repeated MMSE testing are acknowledged. Still, the identification of metabolic-cognitive thresholds as intervention targets is a clinically actionable direction — one that physicians can already act on by prioritizing multimodal prevention in patients showing early cognitive softening alongside metabolic risk.