≡ Menu

New Lyme Culture Test Failed CDC Analysis

lyme disease test revealed as contaminated using DNA analysis of borrelia

Bacterial DNA analysis reveals likely contamination in new Lyme disease test results.

Laboratory contamination has been highlighted as the most likely explanation for the high incidence of positive results for Lyme disease found by researchers developing a new type of test for Lyme disease. Researchers at the US Centers for Disease Control found that almost all of the patients’ samples used as proof of the success of a new Lyme disease culture test contained genetically similar strains of bacteria as those used in developing the test, which would be extremely unlikely to occur naturally.

A Lyme Disease Test with Results Too Good to be True?

The Lyme disease culture test was being developed by Dr. Eva Sapi and colleagues in Connecticut, who published a paper earlier this year outlining their success. The technique involved taking samples of serum from patients known to be positive for Lyme disease bacteria and using cultivating these samples to then test for the presence of Borrelia. Positive cultures were found in 47% of cases within six days of sampling, and in 94% of cases within 16 weeks. A control group of 48 people without Lyme disease was used in the study, comparing these to an initial group of 50 Lyme disease patients and a second group of 72 confirmed cases. The bacteria were identified using immunostaining, polymerase chain reaction (PCR), and direct DNA sequencing.

Caution Over Such Positive Results

The high success rate of this novel test was received enthusiastically by many Lyme disease patient advocates but scientists working in the field of infectious disease advised caution as the method was new and unconfirmed. Such caution appears wise in retrospect as the latest assessment of the method suggests that the high rate of seropositivity found may have been due to laboratory contamination.

DNA Matching of Lyme Disease Bacteria

When testing for Lyme disease using DNA sequencing it is common to see a variety of genetic signatures as Borrelia reproduce and mutate quite rapidly. Homogeneity amongst the results of such tests is unlikely, unless the test cases were deliberately infected in a controlled environment using a single source of bacteria. What the CDC researchers found in analyzing these results from Sapi’s study was that around 80% of the patients’ sample sequences were indistinguishable from those used in the development of the test. Just 10 of the 51 pyrG sequences identified were different to the investigators’ bacterial strains and the CDC scientists note that while they are unsure as to the relevance of these strains, validation of any new testing method is essential to prevent unnecessary antibiotic treatment following a false positive for Lyme disease.

New Test Finds Lyme Disease Bacterial Strains Not Present in US

Barbara J.B. Johnson, PhD, from the Division of Vector-Borne Disease in Fort Collins, Colorado, and colleagues at the CDC were particularly interested in working out why the DNA sequencing used in this study appeared to show that most patients were infected with strains of the Lyme disease bacteria that had not previously been detected in US or Canadian patients. These strains, Borrelia afzelii and Borrelia garinii, are more commonly found in Europe and Asia but are not known to be present in North America.

How the Test was Devised

The researchers developing the test used two Lyme disease bacterial strains commonly employed in laboratory tests for the disease in the US, B31 and 297, alongside the two Eurasian strains. These strains, when found in US patients, are assumed to be the result of exposure to ticks when travelling outside of the US, with no confirmed detection of these bacteria in ticks tested on US soil.

Bacterial Contamination and False Positives

As such, the CDC scientists began to wonder if the high degree of homogeneity between test samples and the presence of unlikely bacterial strains was not a result of exposure and infection but of accidental contamination in the laboratory from the testing protocol itself. Even the different strains of these bacteria would, if infection had arisen naturally, result in different DNA sequences amongst the patients’ samples and so the scientists compared the pyrG sequences for 51 patient isolates using the same primers as those used by Sapi, et al.

Almost Exact Matches for References Strains

What they found was the more than half (53%) of the patients’ samples were infected with B. garinii, with 20 of those 27 found to be identical clones to the strain used as a reference in the test’s development. Differences in the other samples were limited to just a single mutated nucleotide in five cases, two differences in one case, and three differences in the remaining case. These kinds of similarities are highly suggestive of something other than naturally occurring infection.

Less Than 10% of Samples Not Highly Suspect

In addition, 41% of the 51 patients had B. burgdorferi-related sequences and all but one of them had exact matches to the reference strain. Two patients’ samples contained strains of B. afzelii closely related to the reference sample, meaning that 80% (41/51) of the patients’ samples were identical to the test strains, with another 12% differing by a single nucleotide. That leaves just 4 patients whose tests were not an exact match or an almost identical match to the bacterial strains used in this novel, and seemingly highly successful test.

Further Validation Needed

Rather charitably, the CDC researchers note that “further validation of the proposed novel culture method is required,” despite what appears to be a pretty clear indication of something having gone wrong with the testing protocol leading to contamination with the reference strains. The suspected contamination could be due to the use of PCR methods which were employed despite having little merit when the culture contains enough bacteria to be seen under a microscope. As the control samples were not tested using PCR it may be that these avoided the contamination, thus resulting in seronegativity. Had they all been tested by PCR it may be that Sapi, et al, would have found an astonishingly high rate of positivity for Lyme disease amongst those with no other indication of the disease. This, itself, would have (hopefully) sounded the alert about the testing protocols.

Exercising Cautious Optimism Over New Lyme Disease Testing Protocols

Without carrying out the study again, eliminating the PCR testing step, it is impossible to say whether it was this that caused the suspected contamination and, indeed, whether the new test could be helpful in detecting Lyme disease. What the experience does tell us is that a single study presenting results that look too good to be true should lead to cautious optimism and healthy scepticism, rather than raising hopes ready to be dashed.


Barbara J.B. Johnson#, Mark A. Pilgard and Theresa M. Russell, Assessment of New Culture Method to DetectBorrelia species in Serum of Lyme Disease Patients, J Clin Microbiol. Published online August 14, 2013.

Sapi E, Pabbati N, Datar A, Davies EM, Rattelle A, Kuo BA. Improved Culture Conditions for the Growth and Detection of Borrelia from Human Serum. Int J Med Sci 2013; 10(4):362-376. Available from http://www.medsci.org/v10p0362.htm

{ 2 comments… add one }
  • Tom August 29, 2013, 1:00 am

    The CDC has entered their pyrG sequences in the NCBI genbank

    Borrelia burgdorferi 297 CTP synthase (pyrG) gene, partial cds, GenBank: KF170281.1, 707 bp

    Borrelia garinii strain Fuji P1 CTP synthase (pyrG) gene, partial cds, GenBank: KF170282.1, 645 bp

    Borrelia afzelii strain BO23 CTP synthase (pyrG) gene, partial cds, GenBank: KF170280.1, 687 bp

    I used the Virginia.edu DNA sequence comparison tool to do a 1:1 compare of each of the CDC sequences against each of the ALS sequences as in the CDC paper table:

    “Table: Comparison of patient-associated pyrG gene sequences with those of laboratory strains of B. burgdorferi sensu lato.”

    Virginia.edu tool -> http://fasta.bioch.virginia.edu/fasta_w … rm=compare

    The Virginia.edu sequence comparison tool allows one to enter sequences with accession numbers from the NCBI database. It gives a percentage match of aligned sequences and shows differences,

    The CDC paper table percentages are correct for the garinii and azfeli sequences.

    The CDC paper table percentages are in error by one character for the burgdorferi sequences because there is an error in the CDC sequence for KF170281.1 at its character position 618 = “y”. “Y” is not a valid base. This is position 546 in the 603 letter sequence for ALS Bb JX867374.1 where it is a “C”.

    Aside from the CDC error in their sequence, their percentages are correct if one ignores this letter position. The CDC paper should be corrected. They are a 100% match at 602/602, not 603/603. This shows sloppiness by the CDC but isn’t terribly relevant.

    I checked the garinii isolates to see if any of the errors indicated in the CDC table were in the same position. Every error in the garinii isolates appeared random. They all appeared in random positions and letters. I was wondering if they clustered or were random. Random suggests possible sequencing errors while any clusters might imply a unique sequence that simply varied by one letter.

    So there is a troubling match and near matches between the ALS isolates and the ATCC provided B. burgdorferi sensu stricto strain 297 (ATCC 53899), B. afzelii strain BO23 (11) (ATCC76 51992), and B. garinii strain Fuji P1 (11) (ATCC 51991).

    Since garinii has not been found in US patients nor ticks except seabirds, its troubling that 27 of the ALS isolates were garinii. This implies either Borrelia garinii has been missed in the US for over 25 years of research or these people were infected outside the US where garinii is found.

    It is even more troubling that 20 of these have the exact same 603 letter partial pyrG and the other 7 are quite close and appear as random sequencing errors.

    It is further troubling that the ATCC acquired garinii strain Fuji P1 (ATCC 51991) also has the same partial pyrG gene sequence.

    It should be noted that the garinii do also match 2 existing garinii strains from japan and Russia. Not Europe.

    Borrelia garinii strain Ekb704-11 CTP synthase (pyrG) gene, partial cds, GenBank: JX971327.1 Russia

    Borrelia garinii pyrG gene for CTP synthase, partial cds, isolate: HP1, GenBank: AB555778.1 Japan

    I believe ALS needs to explain why a Eurasian garinii strain was found in their study and why they all so closely match and why they also closely match the test ATCC acquired garinii strain Fuji P1 (ATCC 51991).

    I believe this will be difficult to explain unless these people all traveled to Eurasia. Even then it will require follow up studies to truly validate the culture.

    The results from the Borrelia burgdorferi are not as clear cut. There are a number of US strains that match the isolates. The match to the test strain is unfortunate.

    The azfelii result is also not as troubling. Someone in the study might have picked this up outside the US. The match to the test strain is unfortunate.

    The second azfelii looks odd. The nearest azfelii in the NCBI database is a 584/603 match. This is quite distant and does suggest the sequence may have come from a mixed DNA error. Its unclear at best. It would seem odd that no existing NCBI database azfelii strain was even close.

    I’m sorry and not happy to report this as I have been a supporter of the ALS culture and hoped it would be shown to be valid. I am not pro-IDSA or CDC. I have been cultured 3 times positive. I would prefer to learn the culture is valid and confirms my positive serology and would be useful to confirm successful treatment and the end of my infection. This is why cultures are the Gold standard. I also want to know the truth. Invalidating this culture study does not mean anything except just that. Lets hope the follow up studies show the culture is good.

  • Mary Louis September 4, 2013, 4:58 pm

    CDC are guilty of false accusations.
    No contamination could have occurred, as the lab strains were held 200 miles from the lab in which the patients’ blood was cultured.

    CDC also published re DNA sequences they claimed they had filed in Genbank – these were not there. They have now been filed (very late) and discrepancies exist with what they report.

    B. garinii is in Mexico so logically would be in US too.

    Incredibly, Dr McSweegan of NIH pretneded to be a patient for years to fool people into believing the test was invalid. He confessed on LYmeNet that “LHCTom” aka Tom Eames was really him. McSweegan was dismissed from his position as Lyme Program Officer at NIH years ago for harassing patient campaigners.

Leave a Comment