Detecting thyroid cancer early and avoiding unnecessary surgery is the potential promise of a new cancer diagnostic test, known as the Afirma® gene expression classifier test, developed by South San Francisco company, Veracyte.

Human thyroid with cancer nodules.
Photo Credit: Dr Jeffrey Norton/NCI

According to the American Society of Cancer (ACS), with 56,540 cases expected this year, thyroid cancer is the fastest increasing cancer in the United States.

The most common symptom of thyroid cancer is a lump or nodule in the neck. To diagnose cancer, cells are taken by a fine needle from the nodule and examined to see if they are malignant (cancerous) or benign (not cancerous).

The market for the Afirma® diagnostic test from Veracyte is the 15 to 30% of fine-needle aspirations of thyroid nodules that yield indeterminate cytologic findings i.e. you can’t tell if cancerous cells are present or not.

Faced with the uncertainty of “indeterminate” findings, many people elect to have surgery to remove the nodules or thyroid gland. Often the surgery shows that the cells were benign after all.

According to Veracyte, the Afirma® gene-expression classifier test “evaluates the expression patterns of 142 genes to classify indeterminate thyroid nodule FNA samples as benign or suspicious for cancer.”  In other words, it takes the “indeterminate” samples and helps further identify those that do have cancerous cells from those that don’t.

How accurate is the Afirma® gene-expression classifier test?

The New England Journal of Medicine reported earlier this week (online first) in an original article, “Preoperative Diagnosis of Benign Thyroid Nodules with Indeterminate Cytology“, the results of a clinical trial sponsored by Veracyte with 3789 patients at 49 clinical sites.

The paper by Erik Alexander and research colleagues from Brigham and Women’s Hospital and Harvard Medical School Boston, showed that the Afirma® gene-expression classifier test had a 92% sensitivity.  It correctly identified malignancy in 78 out of 85 suspicious nodules (95% confidence interval 84 to 97). The specificity of the test was, however, only 52% (95% CI, 44 to 59).

Details of the trial and the in-depth results can be found in the NEJM paper (open access). Veracyte have also produced a short video, available on YouTube, that discusses the New England Journal of Medicine results:

What do the NEJM clinical trial results mean?

In looking at the efficacy of diagnostic tests, it’s important to understand the distinction between sensitivity and specificity.  Further information on this can be found in an open access paper on Clinical tests: sensitivity and specificity authored by Abdul Lalkhen and Anthony McCluskey. Using the definitions they provide:

Sensitivity = ability of the test to identify those with the disease, in this case those with thyroid nodules that contain cancerous cells.  92% sensitivity means that 92 out of 100 people with an indeterminate thyroid nodule that has malignant cells will have the cancer detected, but 8 will be missed. So the Afirma® test can’t be relied upon to pick up every case of cancerous cells in the 15-30% of people who have nodules with indeterminate cytology.

Specificity = ability of the test to correctly identify those patients without the disease.  A test such as Afirma® with 52% specificity correctly reports 52% of patients without the disease as negative (true negative), but 48% of the patients without the disease are incorrectly labeled as test positive (false positive) i.e. you have evidence of cancerous cells when you don’t.

A diagnostic test that shows high sensitivity and low specificity means that many patients will be told they have cancerous cells, when in fact they don’t. This will subject them to potentially unnecessary treatment or surgery. The test may also miss a small number of people who are told they don’t have cancer, when in fact they do.

The willingness to tolerate the risk from a diagnostic test of possible undetected cancer or misdiagnosed cancer is one for each patient to consider and discuss with their doctor.

What are the implications of this test on clinical practice?

An insightful editorial by J. Larry Jameson, MD PhD, Professor of Medicine and Dean of the Perelman School of Medicine at the University of Pennsylvania accompanies the NEJM trial results, and puts the data into context for clinical practice.

“If results of the gene-expression classifier test were used to inform clinical decision making, it might be possible to reduce surgery for nodules with indeterminate cytology by at least one third, or about 25,000 operations per year,”

Jameson says in his editorial on Minimizing Unnecessary Surgery for Thyroid Nodules. He goes on to note that this would represent “substantial cost savings” and would “reduce unnecessary surgery.” There is also a caveat according to Jameson:

 “The risk of this approach is that 5 to 10% of nodules classified as benign (false negatives) are likely to be malignant, particularly those that are cytologically indeterminate but suggestive of cancer.”

The risk of a false negative (classified as benign but in fact malignant i.e. you do have cancer) suggests the need for repeat testing and monitoring of those at high risk:

“For patients being monitored, it will be important to have a low threshold to repeat fine-needle aspiration if ultrasonographic findings indicate rapid growth or characteristics suggestive of cancer.”

As for those patients who receive a false positive (classified as malignant when in fact benign i.e. you don’t have cancer), no mention is made of the potential need to confirm the diagnosis before treatment such as surgery is undertaken.  While it’s possible that these “indeterminate” patients might have had surgery anyway, the issue of how to deal with false positives could have merited some discussion.

Dean Jameson’s conclusion is that, “this new gene-expression classifier test is a welcome addition to the tools available for informed decision making about the management of thyroid nodules.”

However, reimbursement is key to success for any diagnostic. There’s no point in having an expensive test, no matter how good it may be, if insurance companies won’t pay for it.

How much does the Afirma® diagnostic test cost and is it reimbursed by insurance companies?

Bonnie Anderson, Veracyte’s cofounder and CEO in response to my inquiry about the cost of the test, replied by email that:

“The list price is approximately $4,200, which is a fraction of the cost of a thyroid surgery (approximately $10,000 to $15,000).  The test is covered for all Medicare patients and is being reimbursed by many insurance companies on a case-by-case basis at negotiated rates.  The company has a program in place to ensure that patients do not incur out-of-pocket costs in cases where an insurance carrier does not pay.”

At $4,200 a go, Veracyte believe the test is cost-effective. In support of this the company offers a published cost-effectiveness analysis from Johns Hopkins that was undertaken through a Veracyte research grant. It is beyond the scope of this post to review this.

However, it will be interesting to see whether the test is reimbursed in Europe, and how e.g. the Diagnostics Advisory Committee of the UK National Institute of Clinical Effectiveness (NICE) views the Johns Hopkins cost-effectiveness analysis.

Veracyte partners with Genzyme for promotion and marketing

Earlier this year, Veracyte announced a co-promotion partnership with Genzyme (sanofi-aventis), under which Genzyme would market and promote the test in the United States and globally.

The sales and marketing muscle of a large company suggests that sanofi-aventis believe there is a global market opportunity.

Other companies with diagnostics in development will be closely watching the extent to which the Afirma® gene-expression classifier test is adopted in clinical practice.

A diagnostic test that offers the promise of improved cancer detection in thyroid nodules with indeterminate cytology, and a consequent reduction in unnecessary thyroid surgery is worth paying for in my opinion.

In the future, could gene sequencing make diagnostic tests unnecessary if we are able to use next generation sequencing to identify possible aberrations that are linked to cancer by means of a multi-panel assay when the patient enters the hospital?

The world of gene sequencing and bioinformatics is fast moving and I expect we will see rapid progress over the next few years. The interview on Pharma Strategy Blog with Dr Razelle Kurzrock from MD Anderson is well worth reading if you are interested in this area.