• • The UHPLC-CAD fingerprint method resolved eight major C21 steroidal components with baseline separation; average peak areas ranked tenacissoside A (1.7245) > marsdenoside H (1.2959) > marsdenoside K (0.7517) > tenacissoside B (0.6938) > tenacissoside H (0.6526) > tenacissoside D (0.5413) > tenacissoside I (0.3311) > tenacissoside E (0.2545). This quantitative hierarchy identifies tenacissoside A and marsdenoside H as dominant species, enabling manufacturers to prioritize these markers for raw material screening and batch consistency, reducing reliance on the single pharmacopoeial marker tenacissoside H.
• • Similarity between the reference Tong-guan-teng fingerprint and 12 commercial batches ranged from 0.927 to 0.988, while M. tenacissima stems showed no similarity and lacked all eight C21 steroidal components. This >0.85 threshold provides a validated cutoff for authenticating M. cavaleriei versus M. tenacissima, directly impacting sourcing decisions and preventing adulteration in the supply chain.
• • The method exhibited satisfactory precision, stability, and repeatability, with a total peak area fluctuation of less than 5% across S1–S16 samples, confirming batch-to-batch quality uniformity. This low variability supports the adoption of the semi-quantitative fingerprint as a routine quality control specification, reducing the need for exhaustive absolute quantification.
• • Four abundant components—tenacissoside A, marsdenoside H, marsdenoside K, and tenacissoside B—have reported antitumor activity via modulation of the Akt/GSK-3β/STAT3 signaling axis. Their high relative contents (peak areas >0.69) suggest that a multi-marker quality standard could better predict clinical efficacy than the current single-marker approach, with direct implications for oncology drug development and pharmacopoeial revision.