Download The Clinical Management and Outcome of Nail Salon–Acquired

MAJOR ARTICLE
The Clinical Management and Outcome of
Nail Salon–Acquired Mycobacterium fortuitum
Skin Infection
Kevin L. Winthrop,1,2 Kim Albridge,a David South,a Peggy Albrecht,4 Marcy Abrams,3 Michael C. Samuel,2
Wendy Leonard,3 Joanna Wagner,2 and Duc J. Vugia2
1
Epidemic Intelligence Service, Epidemiology Program Office, Centers for Disease Control and Prevention and 2Division of Communicable
Disease Control, California Department of Health Services, Berkeley, and 3Santa Cruz County Health Agency and 4Santa Cruz Medical Clinic,
Santa Cruz, California
Nontuberculous mycobacterial infections are becoming more common. Recently, Mycobacterium fortuitum
and other rapidly growing mycobacteria have been found to cause severe skin and soft-tissue infections in
association with nail salon whirlpool footbaths. We recently investigated a large outbreak of M. fortuitum
furunculosis among women who received pedicures at a single nail salon. To better define the clinical course
of such infections, we collected clinical details from physicians who were treating outbreak patients. We
constructed multivariable linear models to evaluate the effect of antibiotic treatment on disease duration.
Sixty-one patients were included in the investigation. The mean disease duration was 170 days (range, 41–
336 days). Forty-eight persons received antibiotic therapy for a median period of 4 months (range, 1–6 months),
and 13 persons were untreated. Isolates were most susceptible to ciprofloxacin and minocycline. Early administration of therapy was associated with shorter duration of disease only in persons with multiple boils
(P ! .01). One untreated, healthy patient had lymphatic disease dissemination.
Mycobacterium fortuitum is one of several rapidly growing mycobacteria that are ubiquitous in soil and water
habitats [1–5]. These mycobacteria are known to cause
cutaneous infection, typically in association with trauma or clinical procedures [6]. In 2002, however, Winthrop et al. [7] documented a large outbreak of community-acquired infection with rapidly growing mycobacteria. In this outbreak, 1115 patrons from a single
Received 19 June 2003; accepted 18 August 2003; electronically published 8
December 2003.
Presented in part: 40th Annual Meeting of the Infectious Diseases Society of
America, Chicago, IL, 25 October 2002 (abstract 686).
a
nail salon contracted severe, lower-extremity M. fortuitum furunculosis from contaminated whirlpool footbaths used in the salon. Since this time, additional outbreaks and similar sporadic cases of disease have been
reported [7–9], and it has become clear that whirlpool
footbath–associated mycobacterial infections are more
widespread than was previously appreciated. Because
such infections have only recently been recognized,
their natural history and optimal clinical management
have not yet been elucidated. To document the clinical
and diagnostic features of these infections, we observed
the clinical course of a subset of persons affected by
the outbreak of M. fortuitum furunculosis reported by
Winthrop et al. [7].
Private practice, Watsonville, California.
Reprints or correspondence: Dr. Kevin L. Winthrop, Div. of Tuberculosis
Elimination/Field Services, Centers for Disease Control and Prevention at the
California Dept. of Health Services, Rm. 608, 2151 Berkeley Way, Berkeley, CA
94704 ([email protected]).
Clinical Infectious Diseases 2004; 38:38–44
This article is in the public domain, and no copyright is claimed.
1058-4838/2004/3801-0006
38 • CID 2004:38 (1 January) • Winthrop et al.
PATIENTS AND METHODS
In this observational study, we followed a subset of case
patients previously identified by Winthrop et al. [7]
during an initial outbreak investigation conducted
in 2002. To enroll patients in the study, we identified clinicians
who were treating at least 1 culture-confirmed case patient from
the outbreak and asked them to enroll all of their outbreak
case patients (n p 62) into our study. Outbreak case patients
were defined as persons who, after having receiving a pedicure
at the outbreak salon during the period of April through October 2000, developed a persistent soft-tissue infection below
the knee that lasted for 12 weeks and had ⭓1 of the following
features: negative results of a routine bacterial culture, no response to routine antibiotic therapy, or clinical suspicion of
mycobacterial disease by the treating physician [7].
We asked physicians to complete a standardized questionnaire that gathered clinical details for each case patient. Details
included medical history, treatment regimen, duration of clinically evident infection, and disease outcome. We also recorded
results of culture and antibiotic susceptibility tests, as well as
dermatopathological findings from examinations of skin punch
biopsy tissue specimens. Disease duration was defined as the
interval between the day of disease onset (reported by patients
during the initial outbreak investigation) and the day of clinical
resolution (reported by treating physicians). After disease resolution, physicians asked patients to immediately report any
subsequent reactivation of disease. No attempt was made to
modify or influence the duration or choice of treatment among
physicians. Patient follow-up occurred at intervals determined
by the treating physician.
Patients who received 1 antibiotic (monotherapy) or 2 antibiotics concurrently (dual-agent therapy) directed against mycobacteria for ⭓2 weeks were considered to be treated. Patients
who received no antibiotic treatment, treatment with antibiotics
typically not known to have in vitro activity against mycobacteria, or treatment with antibiotics with activity against mycobacteria for !2 weeks were considered to be untreated. We
evaluated the independent effect of antibiotic treatment using
a variable that adjusted for the elapsed time between disease
onset and the initiation of treatment. With regard to this delay
in therapy initiation, persons treated with antibiotics naturally
clustered into 3 categories of roughly equal numbers on the
basis of the following natural breakpoints in the data: those
who began treatment 1–40 days, 41–70 days, or ⭓71 days after
infection onset.
All case patients were recommended to undergo skin punch
biopsy for mycobacterial culture. Cultures of biopsy tissue specimens were either positive or negative for mycobacteria or were
not performed. We attempted to control for differences in disease duration in our data that could arise because of varying
levels of experience among physicians in treating these infections. To do so, we created a physician variable and categorized
treating physicians as “experienced” or “other” on the basis of
the number of patients they treated. Three dermatologists (i.e.,
experienced physicians) evaluated 30 (50%) of 61 case patients,
whereas the remaining 31 case patients were evaluated by a
total of 18 other physicians, most of whom were primary care
physicians or infectious disease specialists.
The number of boils was recorded at the time of the patient’s
first presentation to the treating physician. In our analysis, we
explored the relationship between the number of boils and
disease duration in a univariate fashion, retaining boils as a
continuous variable, and then grouped the number of boils
into 4 levels (1, 2–3, 3–4, and ⭓5 boils). The only statistically
important relationship between these 2 variables occurred between 1 and ⭓2 boils. Therefore, this dichotomized variable
construct (i.e., 1 and ⭓2) was retained for further analysis in
our model.
All data were entered into EpiInfo 2000, version 1.1 (Centers
for Disease Control and Prevention). Univariate and stratified
analyses were conducted to identify relationships between disease duration and variables that could potentially influence
disease duration. Comparison of means was done with analysis
of variance, and comparison of continuous variables was performed with univariate linear regression. P values for univariate
analysis were calculated using Student’s t test. Multivariable
linear models were then constructed using the Proc GenMod
procedure in SAS statistical software (SAS Institute), to evaluate
the independent effects of various explanatory factors on the
outcome of disease duration. P values for multivariate analysis
were calculated using the Wald test.
RESULTS
Follow-up information was available for 61 of 62 patients; 60
(98%) were female. Patient age ranged from 13 to 53 years.
No persons were immunocompromised; 3 were pregnant at the
time of infection. Patients had a median number of 2 boils
each (range, 1–20 boils). All lesions were present below the
knee in a distribution that corresponded to the area of leg
exposed to water in the whirlpool footbaths used during the
pedicure procedure. The clinical appearance and natural history
of the lesions in this outbreak were strikingly uniform (figures
1–4). Lesions typically first presented as small papules with a
“spider-bite” appearance. Papules underwent characteristic
progression to large fluctuant boils, often with subsequent ulceration, and eventually healed with scarring.
Forty-eight persons received antibiotic therapy (33 received
dual-agent therapy and 15 received monotherapy). Thirteen
persons were considered to be untreated; 11 received no antibiotics, and 2 were treated for !1 week. No persons required
surgical resection of lesions. All treated and untreated persons
eventually had resolution of disease. One initially untreated,
HIV-negative person was given antibiotic therapy after lym-
M. fortuitum Skin Infection • CID 2004:38 (1 January) • 39
Figure 1.
Photograph of a typical boil lesion (size is specified in centimeters) 13 weeks after onset
phatic dissemination of infection and the development of a
large intrathigh abscess that required drainage. Samples of the
abscess were obtained for culture, and culture results were positive for M. fortuitum.
Persons who received antibiotics were treated for a mean
duration of 4 months (range, 1–6 months). Clinicians most
frequently used ciprofloxacin, doxycycline, or clarithromycin
(table 1). Patients often changed drugs during the course of
treatment because of side effects, cost, or other reasons. For
instance, a person who was receiving monotherapy may have
initially been treated with ciprofloxacin for 3 weeks, switched
to doxycycline for 8 weeks, and then switched to minocycline
for 10 weeks. It is notable that doxycycline or minocycline were
the drugs most frequently administered to persons who received
monotherapy (14 of 15 patients, including 12 who received
only doxycycline or minocycline during their entire treatment
course). Patients who were receiving dual-agent therapy underwent similar changes in drugs, and use of the second antibiotic was often intermittent. Ciprofloxacin was the most frequently employed drug in dual-agent therapy (31 of 33
patients), followed by clarithromycin (20 of 33 patients), and
doxycycline or minocycline (17 of 33 patients).
Thirty-two persons had culture-confirmed disease. Of the
remaining patients, 16 did not undergo punch biopsy for culture, 3 had biopsy tissue specimens that were incorrectly processed and not cultured, and 10 had negative results of cultures
of biopsy tissue specimens. Thirty isolates were identified as
M. fortuitum, and 2 were rapidly growing mycobacteria that
were not identified to the species level. Isolates were generally
susceptible to amikacin, cefoxitin, ciprofloxacin, doxycycline,
40 • CID 2004:38 (1 January) • Winthrop et al.
gentamicin, and minocycline and were resistant to sulfa, clarithromycin, azithromycin, and augmentin (table 2). Only 1
(3%) of the 32 positive biopsy tissue specimens had bacilli
demonstrable on an acid-fast bacilli smear. Ten patients with
culture-confirmed disease also had swabs of wound drainage
submitted for mycobacterial culture, and 7 swab cultures were
positive.
Fifteen dermatopathological reports were submitted by treating physicians. Results of histopathological analysis of the lesions were variable and seemed to be consistent with the age
of the lesion that was sampled by biopsy. Older lesions tended
to demonstrate chronic inflammatory changes with few inflammatory cells present. More-acute lesions displayed granuloma
formation, mixed inflammatory cell infiltrate, and local tissue
destruction. Such inflammatory changes were often localized
in or near the hair follicle. Mycobacteria were not seen in any
of the specimens that underwent pathological analysis.
Treated and untreated persons were similar with respect to
the prevalence of culture-positive disease (50% vs. 61%, respectively) and the proportion who were receiving treatment
from experienced providers (50% vs. 54%, respectively). However, treated persons had a higher number of boils than did
untreated persons (mean, 4.7 vs. 2.3 boils; P p .07 ) and were
significantly more likely to have multiple boils (OR, 3.5; 95%
CI, 0.98–12.48; P p .05).
Clinical resolution and disease onset dates were both available for 56 persons (90%) for whom disease duration could
be calculated. Overall, the mean disease duration was 170 days
(range, 41–336 days). By univariate analysis, disease duration
was similar for treated and untreated groups (mean duration,
Figure 2.
Photograph of lesion progression (size is specified in centimeters) after central drainage 14 weeks after onset
168 vs. 177 days, respectively). Persons with culture-negative
disease and those for whom no culture was performed had a
shorter disease duration than did those with positive culture
results (mean duration, 136 vs. 198 days; P ! .01). Persons
treated by experienced providers had significantly shorter durations of disease (mean duration, 143 vs. 195 days; P ! .01).
Initiation of treatment earlier in the disease course was linearly
and significantly associated with shorter mean disease duration
Figure 3.
(P ! .01). Having a single boil versus multiple lesions was also
associated with a shorter mean disease duration (mean duration, 138 vs. 181 days; P p .04).
Stratified analysis suggested that the effect of treatment might
have differed according to the number of boils present. In
persons with multiple boils (n p 41 ), therapy showed a trend
toward shorter disease duration, compared with untreated persons (mean duration, 173 vs. 231 days; P p .07). Among per-
Photograph of lesion progression (size is specified in centimeters) at 17 weeks after onset
M. fortuitum Skin Infection • CID 2004:38 (1 January) • 41
Figure 4.
Photograph of boil resolution, with characteristic hyperpigmented macular scar, 10 months after onset
sons with only a single boil, disease durations were similar in
both treated and untreated groups (mean duration, 147 vs. 123
days; P p 0.50). A final, multivariable linear regression model
controlling for the noted effects of culture result, physician
experience, number of boils, and the interaction between the
number of boils and treatment indicated that early initiation
of antibiotic therapy (1–70 days after disease onset) was associated with shorter disease duration only in persons with
multiple boils (P ! .01) (table 3).
DISCUSSION
We observed the clinical course of 61 persons with lowerextremity M. fortuitum furunculosis acquired from whirlpool
footbaths at a nail salon. Of importance, we demonstrated that
these infections can be treated effectively with oral antibiotic
therapy and that surgical resection is not necessary. Our data
suggest that persons with more extensive disease (i.e., ⭓1 boil)
who initiate antibiotic therapy early in the disease course benefit
most from therapy. Although untreated infections can be selflimited in a healthy host, we found that lymphatic dissemination of infection can occur.
Patients who present for evaluation early in their disease
course—in particular, those with multiple boils—appear to be
the best candidates for antibiotic therapy. Because these infections can be self-limited (at least in healthy hosts), initiation
of antibiotic therapy late in the disease course might not significantly enhance immune system response to the infection.
It is unclear why our analysis failed to find benefit associated
42 • CID 2004:38 (1 January) • Winthrop et al.
with antibiotic therapy in persons with a single lesion. Given
that 14 of 19 patients with a solitary lesion also underwent skin
punch biopsy, it might be that removal of infecting organisms
during the biopsy hastened disease resolution without the aid
of antibiotic therapy.
Although surgical resection of lesions has been reported by
other authors in the treatment of cutaneous infection caused
by rapidly growing mycobacteria [8, 10], our experience indicates that oral antibiotic therapy alone may be sufficient for
Table 1.
Antibiotic therapy administered to
patients who were characterized as having received treatment.
Antibiotic
No. (%) of
a
patients treated
(n p 48)
Ciprofloxacin
34 (71)
Doxycycline
22 (46)
Clarithromycin
21 (44)
Minocycline
12 (25)
Azithromycin
7 (15)
TMP-SMX
3 (6)
Levofloxacin
3 (6)
NOTE. The number of patients treated with each antibiotic in the table exceeds the overall number of persons
treated in the study, because many persons were treated
with 11 antibiotic throughout the disease course. TMPSMX, trimethoprim-sulfamethoxazole.
a
See Patients and Methods for the distinction between
treated and untreated patients.
Table 2.
isolates.
Results of antibiotic susceptibility testing of 29
Antibiotic
Intermediately
susceptible
Mode MIC,
Susceptible
or resistant
mg/mL
1
MIC90,
mg/mL
Amikacin
29 (100)
0 (0)
Ciprofloxacin
29 (100)
0 (0)
Minocycline
10 (100)
0 (0)
Cefoxitin
10 (91)
1 (9)
Doxycyline
16 (89)
2 (11)
Gentamicin
9 (82)
2 (18)
4
8
TMP-SMX
512
.06
.5
10
.25
4
.12
.5
10
2
10 (61)
16 (39)
512
Clarithromycin
4 (14)
25 (86)
16
32
Azithromycin
1 (10)
9 (90)
128
128
Erythromycin
1 (9)
11 (91)
32
32
Augmentin
0 (0)
11 (100)
32
32
NOTE. Data are no. (%) of isolates that underwent susceptibility testing,
unless otherwise indicated. TMP-SMX, trimethoprim-sulfamethoxazole.
treating salon-acquired infection caused by M. fortuitum. The
optimal duration of antibiotic therapy is unknown. Our patients were treated for a mean duration of 4 months. Whether
dual-agent therapy is more advantageous than monotherapy is
also unknown. In addition to low numbers of study subjects
and frequent drug switching by patients, our assessment of this
question was further complicated by the fact that more than
one-half of those who were treated with dual-agent therapy
were, at times, being treated with clarithromycin, a drug to
which most of the M. fortuitum isolates ultimately showed resistance. In such cases, those persons were probably effectively
being treated with monotherapy.
Clinicians could consider choosing dual-agent therapy to
possibly prevent acquired drug resistance. Acquired fluoroquinolone resistance has been demonstrated previously with
ciprofloxacin monotherapy for cutaneous M. fortuitum infection [11], and acquired clarithromycin resistance has been
shown in the single-agent treatment of cutaneous infection
caused by M. chelonae (another rapidly growing mycobacterium) [12]. Because most physicians in our study were aware
of these reports, nearly all patients who were receiving monotherapy were treated with doxycycline or minocycline throughout all or most of the treatment course.
The diagnosis of such infections can be challenging without
clinical suspicion of mycobacterial etiology. Although M. fortuitum and related organisms will grow on standard blood agar
culture media, they typically do so only after 3–7 days under
ideal conditions [6, 13]. Therefore, cultures for routine organisms frequently fail to yield these mycobacteria, because they
are typically not held long enough to demonstrate such growth.
Mycobacteria may also be difficult to collect for culture, particularly if swab samples of wounds are used. Other authors
have recommended the use of skin punch biopsies to capture
organisms for culture, and we indeed found this procedure to
be more sensitive than using swab samples of wounds [14].
Finally, acid-fast bacilli smears appear to have little utility for
diagnosis, because only 1 person in our cohort had bacilli demonstrable on smear.
Because our study was observational by design, our ability
to draw firm conclusions regarding the utility of antibiotic
therapy was limited. To determine disease duration, we relied
on the subjective opinion of many different physicians. It is
possible that the subjective assessment of disease resolution or
patient follow-up frequencies may have varied between physicians, depending on disease severity, treatment group, or some
other variables associated with clinical expertise or training. We
attempted to control for this by including a physician variable
in our model. We were also limited in our ability to assess
outcomes other than disease duration. Although all lesions were
noted to result in scarring, we could not assess whether antibiotics mitigated this outcome.
In summary, clinicians should consider rapidly growing mycobacteria in the differential diagnosis of any difficult-to-treat
soft-tissue infection, particularly when the infection is associated with whirlpool footbaths commonly used in nail salons.
Of importance, we have demonstrated that M. fortuitum furunculosis can be treated successfully with oral antibiotic therapy in otherwise healthy persons and that surgical therapy is
not necessary. Our experience suggests that persons—in particular, those with multifocal disease—benefit most from treatment initiated early in the disease course. Although untreated
infection may be self-limited, dissemination of these infections
can occur among healthy individuals, and persons who are
untreated should be observed closely for such complications.
The risk of disease reactivation is unknown, but, at the time
of writing, no persons in this study have reported recurrence
of disease.
Acknowledgments
We thank Dr. Genevieve Ashcom, Dr. James Beckett, Dr.
Lorena Bischoff, Dr. Andrew Calciano, Dr. Mitra Choudri, Dr.
Table 3. Disease duration, by treatment group, in persons with
multiple boils at presentation to the treating physician.
Treatment
a
group, days
No. of
patients
Disease duration,
mean days (range)
P
None
6
227 (124–316)
Reference group
1–40
13
118 (45–317)
!.01
41–70
7
137 (126–206)
!.01
13
189 (113–336)
.44
⭓71
a
b
b
Interval between disease onset and initiation of treatment.
By multivariate modeling.
M. fortuitum Skin Infection • CID 2004:38 (1 January) • 43
Leslie Christie, Dr. Thomas Deetz, Dr. Barbara Drucker, Dr.
Steven Ellis, Dr. Charles Fishman, Dr. Melvin Gorlick, Dr. Leonard Moore, Dr. Daryl Nounnan, Dr. William Richards, Dr. Vaal
Rothman, Dr. Molly Shields, and Dr. Michelle Violich, for their
collaboration in this study; and Dr. Andrea Winquist (Centers
for Disease Control and Prevention; Atlanta, GA), for her
assistance.
7.
8.
9.
10.
References
1. Collins CH, Grange JM, Yates MD. A review: mycobacteria in water.
J Appl Bacteriol 1984; 57:193–211.
2. Fischeder R, Schulze-Robbecke R, Weber A. Occurrence of mycobacteria in drinking water samples. Zentralbl Hyg Umwelmed 1991; 192:
154–8.
3. Covert TC, Rodgers MR, Reyes AL, Stelma GN Jr. Occurrence of nontuberculous mycobacteria in environmental samples. Appl Environ Microbiol 1999; 65:2492–6.
4. Wolinsky E, Rynearson TK. Mycobacteria in soil and their relation to
disease-associated strains. Am Rev Respir Dis 1968; 97:1032–7.
5. Jones RJ, Jenkins DE. Mycobacteria isolated from soil. Can J Microbiol
1965; 11:127–33.
6. Brown BA, Wallace RJ Jr. Infection due to nontuberculous mycobacteria. In: Mandell GL, Bennett JE, Dolan R, eds. Principles and practice
44 • CID 2004:38 (1 January) • Winthrop et al.
11.
12.
13.
14.
of infectious diseases. 5th ed. Philadelphia, PA: Churchill Livingstone,
2000:2630–6.
Winthrop KL, Abrams M, Yakrus M, et al. An outbreak of mycobacterial furunculosis associated with footbaths at a nail salon. N Engl J
Med 2002; 346:1366–71.
Sniezak PJ, Graham BS, Busch HB, et al. Rapidly growing mycobacterial
infections after pedicures. Arch Dermatol 2003; 139:629–34.
Arizona Department of Health Services. Mycobacterium fortuitum furunculosis associated with salon footbaths. Prevention Bulletin 2001;
15:3.
Rappaport W, Dunington G, Norton L, Ladin D, Peterson E, Ballard
J. The surgical management of atypical mycobacterial soft-tissue infections. Surgery 1990; 108:36–9.
Wallace RJ Jr, Bedsole G, Sumter G. Activities of ciprofloxacin and
ofloxacin against rapidly growing mycobacteria with demonstration of
acquired resistance following single-drug therapy. Antimicrob Agents
Chemother 1990; 34:65–70.
Wallace RJ Jr, Tanner D, Brennan PJ, Broan BA. Clinical trial of clarithromycin for cutaneous (disseminated) infection due to Mycobacterium chelonae. Ann Intern Med 1993; 119:482–6.
Wallace RJ Jr. The clinical presentation, diagnosis, and therapy of cutaneous and pulmonary infections due to the rapidly growing mycobacteria, M. fortuitum and M. chelonae. Clin Chest Med 1989; 10:
419–29.
Wallace RJ Jr, Swenson JM, Silcox VA, et al. Spectrum of disease due
to rapidly growing mycobacteria. Rev Infect Dis 1983; 5:657–79.