A drawn enhancement curve does what visual assessment cannot in O-RADS MRI

Design

Retrospective two-centre diagnostic accuracy study, two readers of differing experience, surgical histopathology as reference standard

Population

341 adnexal masses in 321 women aged 18 to 89 resected between 2008 and 2018; 35.2% malignant, 96.2% with a solid component

Primary outcome

Per-lesion sensitivity, specificity and positive predictive value of O-RADS MRI at a score of 4 or above

Effect

Specificity 0.84 (95% CI 0.79 to 0.89) visually vs 0.92 (0.88 to 0.95) with the intensity curve, p = 0.007; PPV 0.77 (0.69 to 0.83) vs 0.90 (0.83 to 0.95), p = 0.002

Three hundred and forty-one adnexal masses in 321 women, all with surgical histopathology, were reviewed at two referral centres by two radiologists of different experience; 35.2% were malignant and 96.2% had some solid component. Scored by conventional visual assessment of enhancement, O-RADS MRI at a threshold of 4 or above gave sensitivity 0.95 (95% CI 0.89 to 0.98), specificity 0.84 (0.79 to 0.89) and positive predictive value 0.77 (0.69 to 0.83). Substituting a semi-quantitative intensity curve — a manually drawn version of the time-intensity curve — raised specificity to 0.92 (0.88 to 0.95, p = 0.007) and positive predictive value to 0.90 (0.83 to 0.95, p = 0.002). Agreement within and between readers was moderate to high, kappa 0.63 to 0.85. On the study's own modelling, the proportion of benign lesions sent to a gynaecological oncologist would have fallen from 16% to 9%.

What improved is the false positive rate, and that is the failure mode that costs patients here. A benign mass scored as suspicious means a woman is referred to a cancer surgeon, consented for a staging operation and often loses an ovary she did not need to lose. Sensitivity did not fall, so the gain is not the usual trade.

The practical objection is time, and it is answerable. Drawing a curve on the solid component adds minutes to a study that already takes half an hour to acquire, and the agreement figures say a second reader will land in the same place. This is a retrospective series from referral centres with surgical confirmation in every case, so the numbers will be optimistic for a general population — but the direction is clear enough to justify trying it prospectively on your own equivocal cases before it is formally adopted.

• Draw the intensity curve on the solid component in any lesion you are about to score O-RADS 4.
• Compare against the outer myometrium as the reference tissue, as the score requires.
• Report the curve type explicitly, so the gynaecology team can see what drove the score.
• Expect the gain to be in specificity — sensitivity was unchanged at 0.95.
• Audit your own O-RADS 4 lesions against histology before changing local policy on this.

The statistics, in plain English

Positive predictive value is the number that matters to the woman being referred: at 0.77, roughly one in four masses called suspicious was benign; at 0.90, about one in ten. That improvement depends on the 35.2% malignancy rate in this surgical series, and in a general gynaecology population where malignancy is far less common the predictive value of both methods would be lower — that is arithmetic, not a flaw in the study. The specificity figures, 0.84 against 0.92, are the ones that transfer between populations. Note too that this is a retrospective review with histology available for every case, which excludes lesions managed conservatively and so removes the easiest benign masses from the denominator.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.