Wednesday, September 28, 2011

A flawed study claiming prevention of Lyme spirochete infection with topical antibiotics

Two recent papers tested the effectiveness of topical antibiotics in preventing Borrelia burgdorferi infection in mice following a tick bite.  Infection by the Lyme disease spirochete was successfully halted in the Knauer et al. study from Germany1 but not in the Wormser et al. study conducted in New York.2  However a flaw in the Knauer study may have unfairly tipped the outcome in the antbiotic's favor.  (I'll save the Wormser study for another post.)

ResearchBlogging.orgThe paper by Knauer and colleagues1 presented two trials, which differed in how the mice were inoculated with B. burgdorferi,  In the first trial the spirochetes were injected into the skin, and azithromycin was applied topically one hour, three days, and five days later at the injection site.  In the second trial infected ticks transmitted the spirochetes to the mice.   Azithromycin was applied topically to the feeding site immediately after the ticks stopped feeding.  In both trials azithromycin was dissolved in ethanol for application to the inoculation site.  Disseminated infection of the mice was assessed by culturing the heart, bladder, ear, and tarsus 56 days after inoculation.

The results from the first trial reveals the problem with the study (Table 1).  Among the ten mice in the placebo group (first row), which received only ethanol, only one (10%) had any culture positive organs 56 days later.  The spirochetes failed to establish a persistent infection in the other nine mice, suggesting that the investigators were working with a weakened strain of B. burgdorferi.  The ethanol could have had a slight killing effect (see the second trial) yet could not have accounted for the poor infection rate in the placebo group.

The table legend claims that the difference between the placebo and treatment groups was significant, but the statistics were done on the numbers in the column labeled "Infection Status."  According to the text of the paper, a positive "Infection Status" refers to those animals that managed to produce antibodies against B. burgdorferi antigens.  Infection status is therefore not the proper metric to assess the infectivity of B. burgdorferi.  When the statistics are performed with the appropriate numbers, located in the column under "Culture," the effect of azithromycin (1/10 culture positive in control group vs. 0/10 culture positive in any treatment group) is not significant (P = 0.9 for control vs. any treatment group).

Results from the second trial are shown in Table 2.  This trial included an extra control group ("No treatment") that received neither antibiotic nor ethanol.  Four of the seven mice in the untreated group (57%) ended up culture positive.  This is still a low infection rate compared to the rates observed in other studies, in which 90-100% of the control animals end up infected following tick inoculation of B. burgdorferi.  Two of the nine mice treated with ethanol alone (placebo) were culture positive, suggesting that ethanol alone helps prevent infection (57% culture positive in "no treatment" group vs. 22% in placebo group, P = 0.152), although the experiment would need to be repeated with larger groups of animals to make a statistically convincing case.

None of the animals treated with azithromycin were culture positive.  However the number of animals was again too low to conclude that antibiotic treatment was effective (2/9 culture positive in placebo group vs. 0/9, 0/8, and 0/5 in the treatment groups, P > 0.4 for comparison of each treatment group with placebo group).  The authors were able to claim statistical significance by combining the two control groups (no treatment and placebo) and the treatment groups.  However it is inappropriate to combine groups in this manner to attain statistical significance.

Even if the investigators had used a larger number of animals, the problem of their weakened challenge strain remains.  Application of topical antibiotics may turn out to be effective in preventing Lyme disease after a tick bite, but the study presented by Knauer and colleagues was not a fair test of this treatment approach.


1. Knauer, J., Krupka, I., Fueldner, C., Lehmann, J., & Straubinger, R. (2011). Evaluation of the preventive capacities of a topically applied azithromycin formulation against Lyme borreliosis in a murine model Journal of Antimicrobial Chemotherapy DOI: 10.1093/jac/dkr371

2. Wormser, G., Daniels, T., Bittker, S., Cooper, D., Wang, G., & Pavia, C. (2011). Failure of topical antibiotics to prevent disseminated Borrelia burgdorferi infection following a tick bite in C3H/HeJ mice Journal of Infectious Diseases DOI: 10.1093/infdis/jir382


  1. Thank you for posting about this study. When I first read an article which mentioned it, the full text was not yet available online via Pubmed.

    Now that I've read what you've posted and the full text, I am looking at it and wondering about which strain of Bb was in the ticks that fed on mice. They state the injection inoculated mice were given low passage N40, but there is no indication what the ticks had. It makes a difference.

    I find it odd, too, that there weren't more infected mice in the placebo group.

    Two questions:
    You had a post in my RSS reader that was about tick antifreeze but it disappeared. What happened to it?

    How soon will you write about the Wormser topical abx study? I have questions about that one.

  2. I'm perplexed that the study even got published without a basic description of some of the key materials used in the experiments. As you mentioned, there's the question of which Bb strain they used for the tick inoculations. There's also the question of which ticks they used. I assume it was Ixodes, but which Ixodes species? I. scapularis? I. ricinus?

    That tick antifreeze post was accidental. I plan to post the final version within the next few days. I hope to post something about the Wormser study by early next week.