Predicitive value of BARD1 ovarian cancer test

Has anybody looked at the predictive value of the BARD1 ovarian cancer test?

I've looked on the BD1 website and on this forum, but I can't see any discussion about predictive values.

Predictive values depend on sensitivity and specificity of the test and also on the prevalence of a disease in the population.

The statistical method to calculate predictive values is here:

http://gim.unmc.edu/dxtests/Default.htm

Sensitivity and specificity

https://www.bard1.com/pipeline/bard1-autoantibody-tests/

Sensitivityof BARD1 ovarian cancer testis 89%

Specificity of BARD1 ovarian cancer testis 97%

Population incidence of disease

https://www.bard1.com/about/cancer-diagnostics-market/

Take the USA, because it's the main potential market.

22,500 diagnoses p.a.

How many females have ovarian cancer at any given time?

Assume 10 year survival from onset, whether diagnosed or not, plus another 20% undiagnosed beforedeath

No of females having ovarian cancer at any time = 22,500 * 10 * 1.2 = 270,000

No of females in the USA = 165 million

Using the 89% sensitivity, of the 270,000 with the disease, 240,300 will correctly test positive and 29,700 will be false negatives.

Using the 97% specificity, of the 164,730,000 who don’t have the disease, 159,788,100 will correctly test negative and 4,941,900 will be false positives.

So there are 5,182,200 total positives, of which 240,300 are true positives and 4,941,900 are false positives.

So the predictive value of a positive test result is 240,000 / 5,182,200 = 4.64%

i.e. if a test result is positive, there's still only a 4.64% chance that the patient has the disease.

If the 270,000 assumed to have ovarian cancer at any given time is a overestimate, then the predictive value is even less than 4.64%

What are the public health implications?

Is the next step to do biopsies on the 5,182,200 with positive test results in order to find the 4.64% who actually have the cancer?

I don’t know the answer.

The predictive value of a negative test is 159,788,100 true negative / 159,817,800 total negative = 99.98% which is excellent.

It's the low predictive value of a positive test which is the problem IMO.

If I've made any arithmetic or logical errors, could you please point them out.

I picked ovarian cancer because it’s the most advanced in the BARD1 series development pipeline (i.e. not counting the Sienna ones).

II

Disclosures: Non-holder. I do hold other biotechs not in the diagnostic testing area. I have zero medical qualifications. I do have some statistical knowledge. I heard of BD1 because I was looking at Merchant Opportunity Fund's overall investments.