January 13th, 2020

Diagnostic Tests for Syphilis Continue to Perplex Even the Experts: An Unanswerable Question in Infectious Diseases

Here’s a tricky clinical scenario:

  1. An elderly person with cognitive decline or some other non-specific neurologic symptom sees a clinician.
  2. Clinician sends a syphilis screen with a T. pallidum enzyme immunoassay (TP-EIA), which returns positive.
  3. Lab runs a confirmatory test — a T. pallidum particle agglutination test (TP-PA), or similar, which also returns positive.
  4. The lab then runs a rapid plasma reagin (RPR) test, which returns negative.
  5. There is no known prior clinical or lab history of syphilis exposure, diagnosis, or treatment.

Now what are we supposed to do?

I bring this up because Dr. Thomas Fekete raised just this issue on the IDSA’s ID Exchange message board, generating a spirited discussion.

(I’m citing this discussion and quoting with his permission.)

And every card-carrying ID doctor has been asked what to do in just this setting numerous times since labs started using TP-EIA — and not RPR — for syphilis screening a bit over a decade ago. In one study, this pattern (two positive treponemal tests and a negative RPR) occurred in approximately 3% of individuals undergoing testing.

Here’s why the next step is so controversial:  This exact pattern describes three separate clinical scenarios, each of them requiring very different next steps. Here are the potential interpretations:

  1. The result is consistent with, but not diagnostic of, neurosyphilis. Recommendation: Perform a CSF exam to rule out neurosyphilis. This is hardly a trivial undertaking, especially in the elderly. Further complicating this next step is that the CSF exam is notoriously poor at either ruling in or ruling out neurosyphilis. Plenty of false-positives and false-negatives.
  2. The result suggests late-latent syphilis with an RPR that has reverted spontaneously to negative. Recommendation:  Treat with benzathine penicillin 2.4 million units by intramuscular injection, weekly for 3 doses. In this interpretation, clinicians must consider the likelihood of clinical neurosyphilis to be sufficiently low that this result is unrelated to the neurologic symptoms — which begs the question, why was the test ordered in the first place?
  3. The result demonstrates prior treated syphilis, with an adequate serologic response. Recommendation:  No treatment or further testing necessary. Lots of antibiotics have activity against T. pallidum, so antibiotics administered for other indications over the years have inadvertently provided sufficient treatment.

Let’s add to the quandary by quoting Dr. Fekete on two key points:

I cannot find modern information about the incidence of true tertiary or neurosyphilis in elderly patients with this [testing] profile … These patients would not have come to our attention in the old system for syphilis screening.

Where are the prospective clinical series outlining either actual clinical neurosyphilis — or even CSF abnormalities — in those who have this serologic profile?

Plus, in the pre-TP-EIA era, when we used RPR for screening, neurosyphilis would have been considered “ruled out” unless there was a strong prior probability of this disease (which there hardly ever is).

Sometimes ignorance is bliss!

Want a further wrinkle? Some believe that the recommended treatment for latent syphilis — benzathine penicillin, with its long half-life but low CSF concentrations — adequately treats neurosyphilis as well. The thought here is that the immune system plays a role in clearing the infection, so no need for high CNS concentrations of penicillin — except perhaps in people with immunosuppression, as is seen in untreated HIV.

The data supporting this view (like many other aspects of clinical syphilis) are largely uncontrolled and somewhat dated — but strongly endorsed and frequently cited by advocates nonetheless.

But this position is vociferously challenged by others — again with largely anecdotal and outdated data. This group, now in the majority, inform the current CDC guidelines for treatment of neurosyphilis, which recommend high-dose intravenous penicillin G for 10-14 days — a burdensome treatment not easily (or cheaply!) administered to the  elderly, especially those with cognitive impairment.

It’s not as if this neurosyphilis quagmire were a new problem; indeed, the diagnosis of neurosyphilis has been fraught for decades. Often the leading strategy adopted by a hospital or a practice is the one endorsed the most passionately (or most loudly, in case conference) by the local expert or experts.

All this controversy makes this case scenario a classic Unanswerable Question in Infectious Diseases. And perfect for a poll!

So have it at — and educate us by using the comments section to justify your vote.

A 78-year-old man with no known prior history of syphilis or other sexually transmitted infections is evaluated for mild cognitive decline. As part of the work-up, he has the following blood test results:  TP-EIA positive, TP-PA positive, RPR negative.

What would you recommend next?

View Results

27 Responses to “Diagnostic Tests for Syphilis Continue to Perplex Even the Experts: An Unanswerable Question in Infectious Diseases”

  1. Santiago says:

    The probability of a false negative RPR is lower than the probability of a false positive TP-EIA. On top of that, for those considering that this is neurosyphilis, it would then be a neurosyphilis with a false negative RPR… I don’t think that can happen… easily (I’ve learned through time, never to say never in medical cases). So I would consider this a false positive TP-EIA and no treatment indicated.

    What would you think about doing a VDRL instead of an RPR? (why not?)

    • Shishir Gokhale says:

      I agree with your comments. VDRL is more indicative of existing syphilis (treponemal infection) and immune response to it. Serial rise or fall of VDRL test titers also indicate progress or remission. It is unlikely to be false negative. Therefore this test is more valuable.

      The specific serological tests to detect anti treponemal antibodies are more specific but do not differentiate between old/ treated/ partially treated syphilis.

  2. Jeanne Marrazzo says:

    Paul–thanks for distributing this conundrum that perennially confronts us. Responses on the Tom’s initial post were thoughtful and eloquent so not sure I have much to add. I think there is no one right answer–not every person in this scenario has had the same life, exposures, etc and if possible, a good conversation (not to mention exam) with the patient is paramount (I’ve gotten some long-buried stories in this situation–or stories that either not been elicited or attended to). Personally I err on the side of sparing most of these patients the LP and treating for LL syphilis with subsequent clinical follow up if possible. If we had a better CSF diagnostic test to increase the PPV of LP, I could be persuaded to be more aggressive; but most of these LPs will be clean or perhaps have a mildly elevated protein which is hard to ignore once you’ve gone down that pathway.

    I’d add one scenario to your possible three, above. The older literature notes that false positives increase with age not only for non-treponema tests but for treponemal tests as well (largely based on MHT-TP and FTA-ABS, not newer generation tests). More recent studies (see Ina Park’s CID paper last year) cite 100% specificity for TPPA, but I have not seen a true population-based serosurvey that really looked at this in an aging population. This would be an amazing opportunity for NHANES to tackle.

    Happy new year!

  3. Mark Drapkin says:

    I think the key point, which has been mentioned, is that in the early years of syphilis treatment this patient would never have been identified as possibly harboring the illness, since the RPR is negative. Now we are in an era when the pre-LP test probability of late neurosyphilis is much lower, so the likelihood of doing harm by diagnosis/treating seems to dictate doing nothing — assuming physical/neuro exam does not specifically raise suspicion. Hmm..

  4. Stephen Shafran says:

    For decades, no treatment was given when RPR or VDRL was done first and was negative and there is no objective evidence of poor clinical outcomes. CSF exam is invasive, expensive and there is no objective evidence of benefit.

  5. C Chun says:

    There is one more step before the final despair. Repeat RPR by the same method again or by changing to the highly sensitive RPR test. This patient’s reagin level might have been just below the power of the 1st test. Also, it is known that the reagin level could “fluctuate” over time.

  6. Jonathan Underwood says:

    A negative RPR makes neurosyphilis very unlikely so if the prior was low it’s very very unlikely.

    That being said one pragmatic approach is to treat with doxycycline for a month and cover both latent and neurosyphilis and then draw a line under it.

    • Mary Cairns says:

      FM/Psych doc here who works in sub-Saharan Africa (with less-than-reliable laboratory, plenty of HIV, plenty of untreated syphilis at presumably various stages). We don’t have this exact quandary but I have pored over what (little) we know about neurosyphilis and have found it very UN-reassuring for patients of all ages with neuropsychiatric symptoms. I’d love to hear opinions on what “non-specific neurologic symptoms” make people decide a patient is low vs med vs high risk for neurosyphilis and then this weighed against the risks of untreated neurosyphilis (permanent and progressive worsening of such symptoms).

  7. Andrew Whitelaw says:

    From a microbiologist in sub-Saharan Africa – this is a great illustration of the dilemmas we often get presented with. I voted for CSF examination – maybe my lab-based bias and not wanting to ignore a potentially treatable condition. CSF exam will result in 3 outcomes: confidently exclude neurosyphilis; confidently confirm neurosyphilis, and minor abnormalities that neither exclude nor confirm neurosyphilis (probably the most likely outcome – at least anecdotally – so opting for CSF exam may not have been the most logical choice). The other aspect is that ideally one needs a more robust clinical history and exam – how severe is the cognitive decline, are there other more plausible explanations – in which case neurosyphilis becomes even less likely and a CSF exam would not be warranted.

  8. Ruth Mullowney-Agra says:

    So helpful, I saw this patient last week but of course she was “allergic” to penicillin with a reaction the patient could not remember.

    One of the survey responses should (always) be perform a history and physical!

    After she confided that she had been a “player” in her younger days, we opted for a month of doxycycline with the option to LP down the line if she has a more abrupt decline in her mentation

  9. Kevin T Fitzgerald says:

    If negative on everything. Test for Lyme. But everything medical for Lyme is ridiculously faulty. Thats why people get chronically ill. It’s a huge epidemic.

  10. Giuseppe says:

    What about empirical tx with ceftriaxone instead of penicillin? Recommendation on preferring penicillin have quite low quality (like most of the evidences about syphilis) and 2 g of ceftriaxone qd are much easily administered than penicillin q4h or with continuous infusion

  11. Clifford Rosenberg, M.D. says:

    cliffmon@charter.net
    Family Physician retired.
    Sometimes clinical exam can differentiate between different causes of Dementia.
    In my 40years of practice I did not see much tertiary syphilis ( I believe ).
    Are there specific characteristics of lues dementia?

  12. Sanford R. Kimmel says:

    Many, many years ago I did an LP on a patient with positive serological tests for Syphilis. I believe that at the time, this was recommended to exclude neurosyphilis as the patient did not have neurological signs. The LP was negative but the patient ended up with a post-LP headache for about 3 days. After that, I chose to treat similar patients with empiric benzathine penicillin if they were not allergic. Perhaps it was over-treatment but I am not aware of any of the patients developing neuro-syphilis. Of course, this was before the onset of HIV infection.

  13. DC Ahamber says:

    Could this patient have had an undiagnosed syphilis that was inadvertantly treated with a susceptible antibiotic in the past? In cases like this, especially in the setting of cognitive impairment, it would be helpful to corroborate a possible history of treated syphilis or otherwise with the public health department.

  14. Daniel Polowetzky says:

    Is there any role for simply watching and waiting and retesting at a later date?
    Would the labs be expected to change significantly should the patient have either late-latent syphilis or neurosyphilis without posing a meaningful risk to the patient in the meantime?

  15. Helmut Albrecht says:

    I commented on the initial post in ID Exchange, which only further “poked the bear”.
    The good news, this is truly not answerable (at least not evidence based as there is not enough evidence), so there really is not a right or wrong.
    Expert opinion is not as flawed as most people think (The initial discussion is worth reading as some true heavy weights in the field opined) but, in this case, lawyers and the Tuskegee history are more difficult to ignore than in other dilemmas of this kind.
    My solution is to educate providers NOT to draw any syphilis test (specific or RPR) in patients with dementia alone and certainly not in patients with “mild impairment”. Opening the Pandora’s box of LP (guidelines recommend a second LP to document clearance in case of “positives”) is almost always wrong. I have been successful in discussing the flaws of this approach with patient/family and having them make the discussion to spare their loved one this procedure.

  16. Michael Sorokin says:

    From left field: did this man serve or work in the South Pacific at any time – and the serology is the remnant of a Yaws infection

  17. WILLIAM SCHULER says:

    from a simple primary care standpoint, what are the results of serum markers for inflammation in a pt. with neurosyphilis? Or other stages in the elderly, many of whom are still sexuallay active? If you are looking at dangerous or intimidating strategies, would a little ancillary evidence be a help before treating lab tests alone? I do not favor the treatment or diagnostic strategies put forward in a pt. with dementia. What do we hope to accomplish? Reversal of dementia?

  18. RICH KURZ says:

    Do we know that treating NS in elderly with cognitive decline works?

  19. Luiz A. M. Fonseca says:

    I voted for doing a CSF examination to rule out neurosyphillis in view of the patient’s cognitive impairment. I am more familiar with VDRL than with RPR as a guide to monitor syphillis activity and response to treatment, but both tests can have false negative results, although with low probability. And in the public tertiary university hospital where I work, here in São Paulo, Brazil, the cost for the patient is zero and we fortunately have personnel with expertise to both collect the cerebrospinal fluid and perform the tests on it. But anyway, this is always a challenging situation.

  20. steven berman says:

    the reason the test was ordered was ‘cognitive decline’. Now obligated to complete w/u including csf examination

  21. Dr S.Murugan says:

    Dear Paul,
    I am also coming across more discrepancies in serological tests for syphilis reports. TP EIA was positive in one case where as RPR and TPHA came nonreactive. Shall we rule out Syphilis in this case when there was no manifestations?
    Following the recommended treatment in cases of late latent syphilis, RPR remains in the same titre even after two years in many cases, in spite of there was no HIV infection.
    In this case, if you suspect CNS involvement could be due to syphilis (Cognitive impairment), the Injections of Benzathine penicillin is enough to treat a case of neurosyphilis? With Benzthine Penicillin, we can not expect adequate concentration of antimicrobial concentration in CSF to take care of Neurosyphilis? I think we have to admit this patient and give IV Injections of Benzyl Peenicillin G.
    Dr S.Murugan

  22. Matthew Peters says:

    I may be the outlier here, but as a family physician who sees mild cognitive impairment regularly, I am NOT screening all of these patients for syphilis, especially those with low risk sexual histories. I would argue that part of the problem here is overtesting for something that has an incredibly low pretest probability based on the available clinical information. Generally, if I can’t explain their mild cognitive impairment with my H&P and basic labs (usually omitting STI testing unless history warrants), I am more apt to send them for neuropsych evaluation.

  23. Susan Tuddenham says:

    Dear Paul,
    Since this scenario comes up so commonly, we conducted a retrospective chart review of all patients in our hospital system with serodiscordant results (i.e. positive trep and negative nontrep test) who had an LP (Tuddenham et al STD 2015). We could not definitively identify any cases of neurosyphilis (though we did identify a couple of cases of likely ocular syphilis). Wohrl et al. Acta Derm Venereol 2006 was a larger study with similar results. I don’t wish to overstate the importance of our study-it had many limitations and was very far from definitive-though perhaps it provides some (small) reassurance.

    As you point out the data is uncontrolled and dated, however three doses of BPG was a recommended treatment regimen for neurosyphilis through 1982 and results were generally good. Of course, this was before the era of HIV and I would think about immunocompromised/HIV+ patients very differently.

    Every patient has to be considered individually. However, given our (albeit limited) data, the historical efficacy of three doses of BPG, and weighing the potential harms of an LP, in an otherwise immunocompetent, HIV- patient with vague neurologic complaints like the one you describe, I think recommending three doses of BPG with close clinical followup is often reasonable.

  24. Jaan Naktin says:

    General paresis and tabes dorsalis are in no way “cognitive decline” – any reading I do on the late complications of neurosyphilis would suggest that the clinical presentations of these patients is quite severe. Cognitive decline in and of itself, of course, may warrant additional work up including a spinal fluid analysis. In the case as presented, the pre test probability of disease is so poor that the performance characteristics of the testing is very unreliable.

Leave a Reply

Your email address will not be published. Required fields are marked *

Note: This is a moderated forum. By clicking on the "Submit Comment" button below, you agree to abide by the NEJM Journal Watch Terms of Use.

Our physician bloggers cannot respond to requests for personal medical advice, and recommend patients discuss health issues with their individual physicians.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

HIV Information: Author Paul Sax, M.D.

Paul E. Sax, MD

Contributing Editor

NEJM Journal Watch
Infectious Diseases

Biography | Disclosures | Summaries

Learn more about HIV and ID Observations.