Debate Continues as New Studies Support MRI, Not CSF Markers, to Diagnose CJD

Neurology Today
5 June 2007; Volume 7(11); pp 6,10-11

Hurley, Dan

1. ✓ Two new studies aim to improve diagnosis of Creutzfeldt-Jakob Disease (CJD): one found a low sensitivity and specificity for CSF biomarkers for sporadic CJD; another study found MRI an effective diagnostic tool in a small number of patients with familial CJD.

BOSTON-A new, larger study by a San Francisco neurologist who previously challenged the use of the 14-3-3 protein in CSF for diagnosing sporadic Creutzfeldt-Jakob Disease (CJD) has again found a disappointingly low sensitivity and specificity for the biomarker.

Although the specificity and sensitivity of two other biomarkers, neuron specific enolase (NSE) and Total Tau (T-Tau), proved better than those of 14-3-3, neither were good enough to make them useful in diagnosing CJD, concluded Michael D. Geschwind, MD, PhD, assistant professor of neurology at the Memory and Aging Center of the University of California-San Francisco (UCSF). The findings were presented in a poster session here at the AAN annual meeting in May.

MRI appears to be a more reliable diagnostic tool, Dr. Geschwind said, and that view was supported by another poster at the meeting reporting high sensitivity and specificity for familial CJD with diffusion tensor imaging (DTI).

Dr. Geschwind and colleagues examined the sensitivity and specificity for CSF in potential prion cases referred to the UCSF center in the past five years. Among 142 patients tested for 14-3-3, the sensitivity was only 49 percent overall - 47 percent for those with a diagnosis of sporadic CJD, and 50 percent for those with probable CJD.

Among the 40 tested for NSE, sensitivity was 63 percent (64 percent for definite CJD; 60 percent for probable CJD). Among the 20 tested for T-Tau, sensitivity was 70 percent (67 percent for definite CJD; 75 percent for probable CJD).

The specificity of the three biomarkers was 66 percent among the 44 controls tested for 14-3-3, 83 percent among the 18 controls tested for NSE, and 100 percent for the four controls tested for T-Tau.

Our data are contrary to other published results suggesting high sensitivity and specificity of these proteins for sporadic CJD, according to Dr. Geschwind and colleagues. We question the utility of these CSF proteins for CJD diagnosis.

But two neurologists who have led other studies with more positive findings said their views remain unchanged - that the biomarkers are a useful, though imperfect, tool for diagnosing CJD.

The data have limitations but also strengths, said Allen J. Aksamit Jr., MD, associate professor of neurology at the Mayo Clinic College of Medicine in Rochester, MN. If you have a select patient group who fit the clinical criteria for CJD and exclude every other possible diagnosis, it's a fairly sensitive and specific test. You use it as a weight to put on one side for or against the diagnosis.

In 2003, Dr. Aksamit espoused a similar view in an Archives of Neurology editorial (60:803-804) accompanying Dr. Geschwind's first paper challenging the clinical value of the 14-3-3 protein for diagnosis of sporadic CJD. In that study, Dr. Geschwind reviewed 32 pathologically confirmed cases on which 14-3-3 testing had been performed; only 17 cases had a positive result, a sensitivity of only 53 percent (Arch Neurol 2003;60:813-816).

Figure. Dr. Allen J. Aksamit Jr.: The data have limitations but also strengths. If you have a select patient group who fit the clinical criteria for CJD and exclude every other possible diagnosis, it's a fairly sensitive and specific test. You use it as a weight to put on one side for or against the diagnosis.

[ Click here to enlarge ]

The largest data set published so far on biomarkers in CJD patients comes from a European consortium led by Inga Zerr, MD, a neurologist at Georg-August University in Götingen, Germany. She and colleagues reported last year on 1,859 patients with sporadic, genetic, iatrogenic, and variant CJD and 1,179 controls, finding a sensitivity of 85 percent for 14-3-3, 86 percent for T-Tau, and 93 percent when the results of both tests were combined with S100b and NSE (Neurology 2006;67:637-643).

I am aware that the sensitivity of the tests in Dr. Geschwind's laboratory is lower than we reported and I have discussed this with him on several occasions, Dr. Zerr said in an e-mail. I have no idea why the data are so different.

One possible explanation, she and Dr. Aksamit suggested, is that at UCSF - a referral institution where many patients go for a second opinion - Dr. Geschwind is seeing more patients with slowly progressing disease than in the European group. Dr. Zerr's data showed, as have previous studies, that the sensitivity of the biomarkers is highest in patients with the shortest disease duration.

Another possible reason the two studies have reached such different conclusions is that Dr. Zerr included patients with genetic, iatrogenic, and variant CJD, whereas Dr. Geschwind's group includes only sporadic cases. Different case definitions and different testing methods might also contribute to the different findings.

Figure. Dr. Isak Prohovnik: MRI does seem to have better sensitivity and specificity than the biomarkers. But neither our group nor his [Dr. Geschwind's] have sufficient numbers yet.

[ Click here to enlarge ]

In an interview with Neurology Today, Dr. Geschwind said of the European group's findings, I think they may have a selection bias. Every time they publish a study, their sensitivity goes lower and lower. Eventually it will hit ours.

He added in an e-mail, I believe there are occasional cases in which these biomarkers, such as the 14-3-3, total-tau or NSE, may be helpful in diagnosing CJD. The problem is that because these CSF proteins clearly can be elevated in other diseases that clinically mimic CJD, one must be very thorough in ruling out other conditions. Some of these conditions include cancers, neurologic autoimmune (paraneoplastic and non-paraneoplastic) dementias, sarcoid, Hashimoto encephalopathy, and even atypically rapid presentations of other neurodegenerative diseases.

After speaking with Dr. Geschwind in front of his poster, Dr. Aksamit Jr. told Neurology Today: It's all in the selection of the patient population being reviewed, and how high you set the bar for these laboratory tests. If you set the bar high, you'll get higher specificity but lower sensitivity. Our recommendation at Mayo is to set the bar high, to enhance specificity at the expense of sensitivity.

Dr. Aksamit noted, however, that 14-3-3 is no longer being tested at Mayo, because the reagents for a quantitative version of the test, which the institution had preferred over the standard Western blot, are no longer available. Instead, Mayo is testing NSE.

Despite the questions raised by Dr. Geschwind's paper, Dr. Zerr said: I completely agree with Dr. Aksamit that biomarkers in CSF are useful, when tested in the right populations. This is what we always stress when we are talking with physicians: Use CSF in the right differential diagnosis; that is, in dementia.

The one point on which all three neurologists agreed is that MRI seems to be a powerful diagnostic tool. Although MRI is not part of the clinical criteria for CJD yet, said Dr. Zerr, we strongly recommend that neurologists have the MRI done and consider it as an additional test.

Another poster presented at the meeting shared the results of DTI testing on a small group of patients with familial CJD. Isak Prohovnik, PhD, professor of psychiatry and radiology at Mount Sinai School of Medicine in New York City, reported that DTI was abnormal in the basal ganglia of ten of 11 patients, eight of 11 in different cortical areas, and six of 11 in the thalami. No involvement was seen in the brain stem or cerebellum.

When correlated with the clinical findings, DTI was positive for the frontal cortex in eight of nine patients, and for the motor cortex in seven of nine patients. In the basal ganglia, the DTI was positive in 10 patients, although clinical findings were present in only eight. Clinical cerebellar signs were seen in 11 patients and brain-stem signs in seven of the 11, although no positive findings showed up in those regions on either DTI or other MRI sequences.

MRI does seem to have better sensitivity and specificity than the biomarkers, Dr. Prohovnik said, noting that Dr. Geschwind has also used MRI. But neither our group nor his have sufficient numbers yet.

In 2005, Dr. Geschwind co-authored a paper on diffusion-weighted imaging (DWI) and fluid-attenuated inversion recovery (FLAIR) in CJD, finding them to be 91 percent sensitive, 95 percent specific, and 94 percent accurate (Am J Neuroradiol 26:1551-1562). In 2005, Dr. Zerr was the senior author on a paper published in Brain (128:2026-2033), which concluded that FLAIR and DWI improved the clinical diagnosis and should be incorporated in the World Health Organization's diagnostic criteria for sporadic CJD.

• Aksamit AJ. Cerebrospinal fluid 14-3-3 protein: Variability of sporadic Creutzfeldt-Jakob disease, laboratory standards, and quantitation.

• Geschwind MD, Martindale BS, Miller BL, et al. Challenging the clinical utility of the 14-3-3 protein for the diagnosis of sporadic Creutzfeldt-Jakob disease.

• Sanchez-Juan P, Green A, Zerr I, et al. CSF tests in the differential diagnosis of Creutzfeldt-Jakob disease.

• Young GS, Geschwind MD, Dillon WP, et al. Diffusion-weighted and fluid-attenuated inversion recovery imaging in Creutzfeldt-Jakob disease: High sensitivity for diagnosis.

• Tschampa HJ, Kellenberg K, Zerr I, et al. MRI in the diagnosis of sporadic Creutzfeldt-Jakob disease: A study on inter-observer agreement.