Comparison between organismal

staining on histology and tissue culture

in the diagnosis of cutaneous infection:
A retrospective study
Sheila Shaigany, MD,a Alexa Steuer, MPH,b Nicole Seminara, MD,a
Nooshin Brinster, MD,a and Alisa Femia, MDa
New York, New York

Background: In instances of suspected cutaneous infection, the standard of care includes obtaining skin
biopsy specimens for histology and tissue culture. Few studies have compared the clinical utility of each
test.

Objective: To assess the concordance of results between tissue culture and histology, as well as the
clinicopathologic features that may influence the diagnostic yield of each test.

Methods: A retrospective review of all patients who underwent skin biopsy for histology and tissue culture
at New York University from 2013 through 2018.

Results: Of 179 patients, 10% had positive concordance, 21% had positive tissue culture only, and 7% had
positive histology only. We calculated a kappa correlation coefficient of 0.25 between histology and tissue
culture (reference, 0.21-0.39 indicates minimal agreement). Histology exhibited higher sensitivity in
detecting fungi, whereas tissue culture was more sensitive in identifying Gram-negative bacteria.
Antimicrobial use before biopsy led to significantly fewer positive cultures (37.5% vs 71%; P = .023) in
patients ultimately diagnosed with infection.

Limitations: This study was conducted at a single institution, thereby restricting its broad applicability.
The lack of a validated criterion standard to diagnose infection also limits interpretation of the results.

Conclusion: Tissue culture and histopathology often yield discordant results. Dermatologists should
recognize specific limitations, yet high clinical utility in special circumstances, of tests when approaching
cases of suspected infection. ( J Am Acad Dermatol 2020;82:1400-8.)

Key words: bacterial; colonization; concordance; cutaneous infection; deep fungal; diagnosis; histology;
infection; inpatient dermatology; microbiology; mycobacteria; organismal stains; punch biopsy; skin biopsy;
skin infection; special stains; superinfection; tissue culture; ulcer; utility.

P rimary infection of the skin results from a astute dermatologist cannot rely solely on cutaneous
wide range of bacterial, fungal, viral, and morphology or clinical picture to diagnose infection.
parasitic organisms. Systemic infection may In instances in which deeper cutaneous infection is
also present with cutaneous involvement, often as suspected, standard of care includes obtaining skin
the heralding sign of dissemination.1 Even the most biopsy samples for histology and tissue culture.2,3

From the Ronald O. Perelman Department of Dermatology, New Reprint requests: Alisa Femia, MD, Ronald O. Perelman,
York University School of Medicinea; and New York University Department of Dermatology, NYU School of Medicine, 550
School of Medicine.b First Ave, New York, NY 10016. E-mail: Alisa.femia@nyulangone.
Funding sources: None org.
Conflicts of interest: None disclosed. Published online January 28, 2020.
IRB approval status: Reviewed and approved by the New York 0190-9622/$36.00
Univeristy School of Medicine IRB (study #i18-00661_CR1). Ó 2020 by the American Academy of Dermatology, Inc.
Accepted for publication January 22, 2020. https://doi.org/10.1016/j.jaad.2020.01.047

1400
J AM ACAD DERMATOL Shaigany et al 1401
VOLUME 82, NUMBER 6

Although histologic examination and tissue Electronic Data Capture (REDCap) (Vanderbilt
culture are routinely performed in cases of suspected University, Nashville, TN). We collected the
infection, few studies have elucidated their following data: patient demographics, biopsy
concordance or diagnostic yield. Xia et al4 examined setting, lesion morphology, biopsy size, histologic
the concordance of microbiological studies with findings, including special stains, tissue culture
histology in 150 biopsy samples; however, they results, prebiopsy differential diagnosis, final clinical
consolidated organismal stains and cultures into 1 diagnosis, presence of immunosuppression,
category, thereby limiting the evidence of antimicrobial/
ability to draw conclusions antifungal use at any point
regarding individual tests. CAPSULE SUMMARY in the 14 days before biopsy,
Other studies have focused treatment, and treatment
narrowly on subsets of or- d
Skin biopsy for histology and tissue response.
ganisms such as nontubercu- culture are routinely performed in cases Biopsy size was calculated
lous mycobacteria5-7 or of suspected infection. from the surface area of the
mycotic infections8,9 or on d
Both tests show low sensitivity for top portion of the biopsy (ie,
specific populations, such as predicting infection. Dermatologists 4-mm punch = 4p mm2;
immunocompromised indi- should be aware of the diagnostic 3-mm punch = 2.25p mm2;
viduals, children, or patients strengths of histology (ie, fungi) versus 4-mm punch split in half = 2p
with ulcers.10-13 Limitations tissue culture (ie, Gram-negative mm2; 6-mm punch split in
in the existing literature lead bacteria), and the negative impact that half = 4.5p mm2) to account
to challenges in the inter- antimicrobials may have on results. for biopsies split in half for
pretation of test results, hematoxylin-eosin and tissue
especially when discordant, culture. Immunosuppression
and further investigation on the utility and yield of was defined as any patient who was treated with
histology and tissue culture is warranted. immunosuppressive medications (within 60 days
before biopsy) or cytotoxic chemotherapy (within
OBJECTIVES 30 days before biopsy), received a solid organ or
We aimed to illuminate the concordance of results allogenic hematopoietic stem cell transplant, or
between histology and tissue culture from biopsies was diagnosed with hematologic malignancy,
performed on all patients with suspected clinical aplastic anemia, myelodysplastic syndrome, active
infection and to identify patterns in concordant solid tumor malignancy, primary immunodefi-
versus discordant cases, clinical features potentially ciency, HIV/AIDS, cirrhosis, or end-stage renal
affecting test results, and the diagnostic yield of disease.
histopathology versus tissue culture. We classified the histology result as positive if any
organism was visualized on dermatopathology
METHODS (hematoxylin-eosin or special organismal staining)
Patient selection and mentioned in the pathology report. All
After institutional review board approval at New specimens were reviewed by New York University
York University, we performed a retrospective chart dermatopathologists. A subset of biopsies with
review of all patients ages 0 to 90 years seen by positive culture and negative histology results
members of the Department of Dermatology from was independently re-reviewed by a New York
September 18, 2013, through July 6, 2018 who Univeristy dermatopathologist (NB), but biopsies
underwent skin biopsies for both tissue culture and were categorized according to their original read.
histopathology within 3 weeks of one another for the We defined tissue culture result as positive if an
same cutaneous process. organism grew in tissue culture, excluding estab-
lished common contaminants such as coagulase-
Data collection negative staphylococci, Corynebacterium species,
We reviewed 204 patient biopsy results in the Bacillus species, Propionibacterium acnes,
electronic medical record system. Biopsies Viridans group streptococci, Clostridium perfrin-
performed by nondermatologists, nail specimens, gens, Micrococcus species, and enterococci.14
and circumstances in which biopsy and tissue culture Exceptions were made for contaminants that
were performed on separate cutaneous processes correlated with final clinical diagnosis (ie,
were excluded (n = 25). A total of 179 skin biopsy Corynebacterium species in pitted keratolysis).
results from 176 patients met inclusion criteria. The prebiopsy differential diagnosis and final
Deidentified data were recorded on Research clinical diagnosis were divided into 2 categories:
1402 Shaigany et al J AM ACAD DERMATOL
JUNE 2020

Table I. Patient demographics and clinical characteristics in all patients and by subgroup
All Positive histology and Positive histology Positive tissue
Characteristics patients tissue culture result result only culture result only
Total, n 179 17 12 37
Age, mean 6 SD 53 6 18.3 53 6 18.3 55 6 22 58 6 17.9
Female, n (%) 87 (48.6) 7 (41.2) 6 (50) 17 (46)
Inpatient, n (%) 104 (58.1) 13 (76.5) 6 (50) 18 (48.6)
Type of lesion, n (%)
Macule/patch 12 (6.7) 1 (5.9) d d
Papule/plaque 75 (41.9) 5 (29.4) 6 (50) 17 (45.9)
Nodule/subcutaneous 34 (19) 4 (23.5) 3 (25) 2 (5.4)
Exanthem 2 (1.1) d d d
Ulcer 43 (24) 5 (29.4) 3 (25) 17 (45.9)
Pustule 5 (2.8) 2 (11.8) d 1 (2.7)
Abscess 1 (5.6) d d d
Bullae/vesicles 7 (3.9) d d d
Cultures performed, n (%)
Bacterial 171 (95.5) 17 (100) 9 (75) 37 (100)
Fungal 164 (91.6) 16 (94.1) 10 (83.3) 33 (89.2)
AFB 163 (91.1) 16 (94.1) 10 (83.3) 33 (89.2)
Histology stains, n (%)
Gram 89 (49.7) 13 (76.5) 7 (58.3) 18 (48.6)
PASD 134 (74.9) 14 (82.4) 11 (91.7) 26 (70.3)
AFB or Fite 86 (48) 11 (64.7) 7 (58.3) 13 (35.1)
GMS 8 (4.5) d 2 (16.7) d
None 37 (20.7) d 1 (8.3) 10 (27)
Prebiopsy suspicion of infection, n (%) 98 (55.1) 17 (100) 8 (66.7) 16 (43.2)
Final diagnosis, n (%)
Infectious 64 (35.8) 15 (88.2) 9 (75) 17 (45.9)
Primary infection 55 (30.7) 13 (76.5) 8 (66.7) 11 (29.7)
Superinfection 9 (5) 2 (11.8) 1 (8.3) 6 (16.2)
Not infectious* 115 (64.2) 2 (11.8) 3 (25) 20 (54.1)
Neutrophilic dermatoses 21 (18.3) 1 (5.9) 2 (16.7) 4 (10.8)
Venous stasis 5 (4.3) 1 (5.9) d 3 (8.1)
Neoplastic 8 (7) d d d
Vasculitis 12 (10.4) d d 2 (5.4)
Inflammatory 19 (16.5) d d 4 (10.8)
Granulomatous 9 (7.8) d d d
Panniculitis 5 (4.3) d d d
Viral 9 (7.8) d d 4 (10.8)
Drug 7 (6.1) d d d
Other 19 (16.5) d 1 (8.3) 3 (8.1)
Change in diagnosis (n = 165), n (%) 59 (35.8) 2 (11.8) d 1 (2.7)
From infectious to not infectious 48 (29.1) 2 (11.8) d d
From noninfectious to infectious 11 (6.7) d d 1 (2.7)
Biopsy and tissue culture performed 164 (92) 16 (94.2) 12 (100) 33 (89.2)
on same day, n (%)
Lapsed days from biopsy to tissue culture Median = 9 16 d Median = 5.5
(when not performed on same day) Mean = 8.2 (n = 1) Mean = 6.5
(n = 15) (n = 4)
Immunocompromised, n (%) 79 (44) 7 (41.2) 5 (41.7) 15 (40.5)
Antibiotics, n (%) 98 (54.7) 7 (41.2) 7 (58.3) 16 (43.2)

AFB, Acid-fast bacillus; GMS, Grocott-Gomor methenamine silver; PASD, periodic acideSchiffediastase; SD, standard deviation.
*Does not include noninfectious primary diagnoses with superinfection.

infectious and noninfectious. The infectious was defined as clinically suspicious for infection if
category included primary infection and super- infection was listed as one of the top 2 differential
infection and excluded viral infections. A lesion diagnoses before biopsy.
J AM ACAD DERMATOL Shaigany et al 1403
VOLUME 82, NUMBER 6

Fig 1. Tissue culture and histology concordance.

Statistical analysis Skin biopsy with tissue culture resulted in a

Biopsies were divided into 1 of 4 groups: a change in diagnosis (infectious to noninfectious or
negative concordant group, a positive concordant vice versa) in 35.8% of cases. Although 17 (9.5%)
group (positive histology and positive tissue culture samples showed both positive histology and tissue
results), a positive histology only group, and a culture results (positive concordance), 12 (6.7%) had
positive tissue culture only group. Descriptive positive histology result only, and 37 (20.7%) had
analyses were performed across the latter 3 groups. positive tissue culture result only (Fig 1).
A kappa correlation coefficient was calculated to We next examined the spectrum of organisms
assess the concordance of results between histology identified by each test to illuminate if one test is
and tissue culture, as were sensitivities and superior for detecting specific organisms (Table II).
specificities for each test. Statistical comparisons Specimens with positive histology only showed the
were performed by using the chi-square test. highest proportion of deep fungal (3/12), parasitic
Statistical analyses were performed in Stata, version (2/12), or atypical mycobacterial organisms (1/12) on
14.0 (StataCorp, College Station, TX). histology, suggesting that histology is more sensitive
than tissue culture for fungal organisms. In contrast,
the majority of biopsies with positive concordance
RESULTS (14/17) contained bacteria, with Staphylococcus
Table I shows clinical characteristics for 179 aureus accounting for 11 of 14 cases. The positive
patients. Overall, 36% of patients were ultimately tissue culture only group also predominantly repre-
diagnosed with infection. Histology and tissue sented bacterial organisms but exhibited a higher
culture results agreed in 130 (72.6%) biopsies, with presence of Gram-negative bacteria and polymicro-
a calculated kappa correlation coefficient of 0.25 bial cultures, suggesting that tissue culture is more
(reference: 0-0.20, no agreement; 0.21-0.39, minimal; sensitive in detecting Gram-negative bacteria. When
0.40-0.59, weak; 0.60-0.79, moderate; 0.80-0.9, histology slides from 27 of 37 specimens with positive
strong; [0.90, almost perfect).15 tissue culture only were re-reviewed, 10 biopsies
1404 Shaigany et al J AM ACAD DERMATOL
JUNE 2020

Table II. Breakdown of organisms according to subgroup, (n)

Positive results on histology Positive histology results Positive tissue culture results
and tissue culture (n = 17) only (n = 12) only (n = 37)

Gram-positive bacteria (11) Gram-positive bacteria (3) Gram-positive bacteria (17)

- Staphylococcus aureus (10) - Impetiginization (2) - Staphylococcus aureus (18)*,6
- Corynebacterium (1) - Pyoderma gangrenosum (1) - Staphylococcus haemolyticus (1)
- Streptococcus pyogenes (1)*,1
Gram-negative bacteria (3) Gram negative bacteria (1)y - Actinomyces species (1)
- Escherichia coli (1)*,1 - Folliculitis (1)
- Klebsiella species (2)*,z,2 Gram-negative bacteria (20)
- Proteus species (2)*,2 Atypical mycobacteria (1) - Pseudomonas species (9)*,5
- Pseudomonas species (1)*,1 Fungal: deep (3) - Serratia species (5)*,2
- Cryptococcus species (1) - Stenotrophomonas species (1)*,1
Fungal (3) - Scedosporium species (1) - Proteus species (2)*,2
- Trichophyton rubrum (1) - Mucormycosis (1) - Citrobacter species (2)*,2
- Candida krusei (2) - Escherichia coli (4)*,2
Fungal: superficial/follicular (2)
Leishmaniasis (2) - Klebsiella species (2)*,1
Fungal (2)
- Trichophyton tonsurans (1)
- Exophiala species (1)

*Number grown in polymicrobial culture.

y
Gram-negative bacteria seen in conjunction with Gram-positive bacteria in 1 case.
z
Seen in conjunction with positive results for yeast on histology.

showed bacteria not previously reported by the whether the presence of antimicrobials, immuno-
original dermatopathologist, because they were not suppression, or size of biopsy specimen affected
considered to be contributory to the primary histo- the positive yield of tissue culture or histology (Fig
pathologic process. These cases all contained Gram- 2). Among patients taking antimicrobials, we noted
positive cocci residing within the superficial a lower proportion of positive culture results
epidermis or within a hair follicle, and no (37.5%; P = .023) and positive histology results
Gram-negative bacteria were seen, even in patients (32.5%; P = .401) compared with individuals not
with growth of Gram-negative bacteria in tissue taking antimicrobials (70.8% with positive tissue
culture, highlighting the difficulty in visualizing culture results, 45.8% with positive histology re-
Gram-negative organisms on histology. sults). Neither immunosuppression nor size of
On descriptive subgroup analysis, we found that biopsy significantly altered the yield of positive
100% (17/17) of biopsies in the positive concordant histology/tissue culture within this population.
group listed infection as 1 of the top 2 prebiopsy Notably, of patients diagnosed with infection, 51
differential diagnoses, compared with 66.7% (8/12) had follow-up data, and 47 of these 51 (92.2%)
in the positive histology only group and 43% (16/37) patients showed clinical improvement/resolution in
in the positive tissue culture only group (Table I). response to targeted antibacterial/antifungal treat-
Moreover, although 88% (15/17) of patients with ment; the remaining 4 patients died of severe
positive concordance and 75% (9/12) with positive systemic infection (ecthyma gangrenosum, purpura
histology only were ultimately diagnosed with fulminans, and 2 cases of disseminated fungal
cutaneous infection, only 46% (17/37) with positive infection).
tissue culture only were deemed to be clinically Finally, we calculated sensitivities and specific-
infectious. In other words, patients with biopsies ities for each test utilizing a final clinical diagnosis of
having positive histology results only were more infection as the criterion standard to determine true
likely than those with positive tissue culture results positives/negatives (see Table III). Both tissue
only to receive a final clinical diagnosis of infection, culture and histology exhibited sensitivities of 0.50
and the prebiopsy clinical diagnosis was thought to or less. Histology showed a higher specificity (0.96 vs
be infectious in 100% of patients with concordant 0.81) and higher positive predictive value (0.83 vs
histopathology and tissue culture. 0.59), whereas both histology and tissue culture
We also analyzed the 64 patients ultimately showed similar negative predictive values of 0.73
diagnosed with cutaneous infection to determine and 0.74, respectively.
J AM ACAD DERMATOL Shaigany et al 1405
VOLUME 82, NUMBER 6

Fig 2. Effects of antimicrobials, immunosuppression, and size of biopsy specimen on test

results. *A 4-mm punch biopsy is equivalent to having a surface area of 4p mm2. ^Specimens
split in half were categorized according to their surface area (ie, 4-mm punch specimen split in
half = 2p mm2; 6-mm punch specimen split in half = 4.5p mm2) relative to that of a 4-mm punch
biopsy specimen (4p mm2).

DISCUSSION detailed clinical information in concordant versus

Our study is the largest to date to analyze concor- discordant cases; the potential effects of anti-
dance between organismal staining on histology and microbials, immunosuppression, and size of biopsy
tissue culture among all patients seen by dermatolo- on diagnosis of cutaneous infection; and a compari-
gists. Moreover, this study is the first to report son of sensitivities and specificities for each test.
1406 Shaigany et al J AM ACAD DERMATOL
JUNE 2020

Table III. Diagnostic yield of tissue culture versus was previously described in association with deep
histology* cutaneous fungal infection and attributed to a
Measure Calculation Tissue culture Histology
combination of the homogenization of tissue in the
microbiology lab, nonviable or walled-off fungi, the
Sensitivity TP 0.5 0.38
use of preservative containing local anesthesia, and
TP 1 FN
Specificity TN 0.81 0.96 variability in the location of tissue sampling.8 Thus,
TN 1 FP in cases of suspected fungal infection, histology may
PPV TP 0.59 0.83 be more reliable than tissue culture via punch
TP 1 FP biopsy.
NPV TN 0.74 0.73 Another interesting feature of subgroup analysis
TN 1 FN was that notably, except for 1 case of Gram-negative
rods seen around a hair follicle, Gram-negative
FN, False negative; FP, false positive; NPV, negative predictive
organisms were never visualized on histology. Poor
value; PPV, positive predictive value; TN, true negative; TP, true
positive. visualization of Gram-negative organisms on
*True positive was defined as a positive test result plus a final histology likely results from a lack of contrast
clinical diagnosis of infection, including cases of both primary between bacteria and counterstain in the Brown-
cutaneous infection and superinfection. False positive was defined Brenn Gram stain.17,18 Therefore, we stress the
as a positive test result plus final clinical diagnosis of not
importance of obtaining a tissue culture in instances
infectious.
with high suspicion for Gram-negative infection,
Among patients ultimately diagnosed with such as ecthyma gangrenosum, malakoplakia, or
cutaneous infection, we found that the use of purpura fulminans.
antimicrobials at the time of biopsy significantly Additionally, in the majority of biopsies with
decreased the yield of positive tissue culture results positive histology results only, infection was
and, to a lesser extent, positive histology results, ultimately diagnosed, and at a similar rate to those
although the latter observation was not statistically with positive concordance. This observation is
significant. Therefore, we recommend that a skin reflected by the high specificity for histology in
biopsy for histology and tissue culture be obtained predicting cutaneous infection; however, this may
before the initiation of antimicrobials, because even also reflect a potential bias in the dermatologist to
1 dose may negatively affect the yield. Within this hinge final diagnosis on histology rather than tissue
same population, our data showed no appreciable culture.
relationship between immunosuppression and test To contrast, the predominant presence of
outcome. This may be partially attributed to the bacterial organisms identified in the positive
observation that immunocompromised patients are concordant group and positive tissue culture only
frequently started on antimicrobials and are thus group suggests that tissue culture may be more
prone to false negative tissue culture results. useful in the identification of bacterial organisms,
Moreover, despite recommendations that larger although not necessarily relevant ones, given the
biopsy specimens (at least 6-mm punch) be taken high proportion of clinically insignificant bacteria in
for tissue culture,16 we observed no direct the positive tissue culture only group, in which
relationship between size of biopsy and positivity roughly half of all positive culture results arose
of test results within individuals ultimately diagnosed from ulcers in which noninfectious etiologies such
with infection. These findings, however, are limited as pyoderma gangrenosum or venous stasis, were
by a relatively low number (n = 8) of biopsy samples ultimately diagnosed. Therefore, a positive result on
with a surface area greater than 4p mm2 (equivalent tissue culture of an ulcer does not definitively
to a 4-mm punch), which serves to highlight the fact exclude a noninfectious cause and may, in fact,
that dermatologists often perform biopsies smaller represent colonization, a notion further supported
than the recommended size when obtaining tissue by the high frequency of polymicrobial cultures
cultures, thereby constricting the statistical power. detected in this group (10/37 [27%]). This
Close examination of each subgroup showed observation contrasts with Khoobyari et al’s13 recent
several distinct patterns. Of those with positive investigation on the utility of skin biopsy and culture
histology only, a relatively higher proportion of in chronic ulcers, in which they described a higher
fungi were identified, suggesting that histology diagnostic yield for tissue culture and recommended
serves as a more reliable modality for detecting performing tissue culture preferentially over special
fungal infection compared with tissue culture stains on histology in ulcers with suspected
obtained from a punch biopsy. This phenomenon infection.13
J AM ACAD DERMATOL Shaigany et al 1407
VOLUME 82, NUMBER 6

Interestingly, when biopsy slides with positive validation to diagnose cutaneous infection. Thus,
tissue culture only were re-reviewed, more than one calculations are derived from the final clinical
third of biopsy specimens were noted to contain diagnosis, which is inherently influenced by
bacteria that correlated with the organism grown in histology and tissue culture results.
culture (usually S aureus) but, from a histopatholo-
gic perspective, were considered irrelevant to the Limitations
diagnosis, highlighting the subjective nature of We recognize the limitations of our study due to its
histologic results compared with tissue culture. retrospective design and limited sample size,
These findings suggest that it may be helpful for although we emphasize that, to our knowledge, this
dermatopathologists to always report the presence is the largest study to date to assess the concordance
of organisms with notation regarding pertinence to of tissue culture and histopathology results, 2 widely
the primary histologic findings. Bacteria detected on used diagnostic tests, among a broad population of
histologic re-review were all located superficially in patients. We call special attention to the inherent
the epidermis or around the hair follicle and in 3 of limitation that no diagnostic criterion standard for
10 cases were detected in patients in whom cutaneous infection exists, and thus, we are subject to
superinfection was ultimately diagnosed, an entity the limitations of using final clinical diagnosis as the
that we considered to be cutaneous infection criterion standard in this study. We highlight the
because it influenced management. inherent limitation of using final clinical diagnosis in
Our data showed a concordance of 72.6%, this manner, because consideration of organismal and
consistent with existing literature by Xia et al,4 who histologic data is taken into account when rendering
described a 70.7% concordance among 150 skin the final clinical diagnosis. As such, using final clinical
biopsy specimens using different methodology. Our diagnosis as our standard affects the ability to interpret
calculated kappa correlation coefficient of 0.25 our sensitivity/specificity analysis. Clinical improve-
translates to a minimal level of agreement between ment was observed in the vast majority of patients
histology and tissue culture (reference: 0.21-0.39 treated with targeted antimicrobial therapy, which
indicates minimal agreement).15 We regard the may help validate our use of final clinical diagnosis;
kappa correlation coefficient as a more accurate however, it is important to recognize the limitations in
statistical measure of concordance, because it this reasoning because patients may have received
accounts for the possibility of agreement by chance multiple therapies, an alternate diagnosis may have
and does not require a method of validation. The been incidentally treated with chosen therapy, and
observation that an infectious pathogen was some patients may have spontaneously improved
detected in both histology and tissue culture in irrespective of therapy. With these limitations in mind,
only 23.4% (15/64) of patients with a diagnosis of we believe that using final clinical diagnosis as the
infection highlights the high frequency of discordant standard has clinical utility, not only for the
results and the need to assess individual test utility interpretation of tissue culture and histopathologic
relative to the type of organism suspected of results but also for understanding potential biases of
causing infection (ie, Gram-positive bacteria vs the clinician in arriving at a final diagnosis.
Gram-negative bacteria versus fungi).
Finally, our analysis on the diagnostic yield of CONCLUSION
histology and tissue culture showed an overall low We found an overall low concordance between the
sensitivity (#50%) for both tests yet high specificity results tissue culture and histology, 2 tests frequently
(96%) for histology. Xien et al4 reported a ordered concomitantly in cases of suspected
higher sensitivity (80.8%) yet lower specificity cutaneous infection. Neither test was sensitive in
(59.2%) for microbiological studies (which included identifying cutaneous infection. Histology showed a
tissue cultures and special stains) and did not higher specificity and positive predictive value
separately describe sensitivities or specificities for overall, in particular for fungal infection, whereas
histology.4 Our findings of tissue culture sensitivity tissue culture played a more pivotal role in the
of 50% and specificity of 81% were more closely identification of bacterial organisms, especially
aligned with Chen et al’s10 study of patients with Gram-negative bacteria. Caution must be used in the
cancer with rash, which reported a tissue culture interpretation of tissue cultures of ulcers, because
sensitivity of 25% and specificity of 93%.10 Of note, in noninfectious ulcers may exhibit polymicrobial
contrast to the kappa correlation, the sensitivity and bacterial growth, representing colonization. Finally,
specificity values presented in our study and although the use of antimicrobials seems to affect the
previously mentioned studies are statistically flawed, yield of tissue culture and histology, the size of biopsy
because there is no independent method of and immunosuppression do not.
1408 Shaigany et al J AM ACAD DERMATOL
JUNE 2020

REFERENCES 9. Roy P. Revisiting the utility of histopathological examination

1. Wolfson JS, Sober AJ, Rubin RH. Dermatologic manifestations of biopsy: a necessity in microbiology. JCDR. 2017;11(5):DC16-
of infections in immunocompromised patients. Medicine. DC18.
1985;64(2):115-133. 10. Chren MM, Lazarus HM, Salata RA, Landefeld CS. Cultures
2. Wolfson JS, Sober AJ, Rubin RH. Dermatologic manifestations of skin biopsy tissue from immunocompromised patients with
of infection in the compromised host. Annu Rev Med. 1983;34: cancer and rashes. Arch Dermatol. 1995;131(5):552-555.
205-217. 11. Chren MM, Lazarus HM, Bickers DR, Landefeld CS. Rashes in
3. Stevens DL, Bisno AL, Chambers HF, et al. Practice guidelines immunocompromised cancer patients. The diagnostic yield of
for the diagnosis and management of skin and soft tissue skin biopsy and its effects on therapy. Arch Dermatol. 1993;
infections: 2014 update by the Infectious Diseases Society of 129(2):175-181.
America. Clin Infect Dis. 2014;59(2):147-159. 12. Smith RJ, Klieger SB, Sulieman SE, Berger E, Treat JR, Fisher BT.
4. Xia FD, Song P, Joyce C, Mostaghimi A. The utility of Retrospective review of immunocompromised children
microbiological studies in diagnosis and management of undergoing skin biopsy for suspected invasive infection:
suspected dermatological infection. JAMA Dermatol. 2017; analysis of factors predictive of invasive mold. Pediatr
153(11):1190-1192. Dermatol. 2017;35(1):104-111.
5. Laga AC, Milner DA Jr, Granter SR. Utility of acid-fast 13. Khoobyari S, Miller TI, Shinohara MM. Utility of skin biopsy and
staining for detection of mycobacteria in cutaneous gran- culture in the diagnosis and classification of chronic ulcers: a
ulomatous tissue reactions. Am J Clin Pathol. 2014;141(4): single institution, retrospective study. Am J Dermatopathol.
584-586. 2019;41(5):343-346.
6. Dodiuk-Gad R, Dyachenko P, Ziv M, et al. Nontuberculous 14. Hall KK, Lyman JA. Updated review of blood culture
mycobacterial infections of the skin: a retrospective study of contamination. Clin Microbiol Rev. 2006;19:788-802.
25 cases. J Am Acad Dermatol. 2007;57:413-420. 15. McHugh ML. Interrater reliability: the kappa statistic. Biochem
7. Bartralot R, Pujol RM, Garcia-Patos V, et al. Cutaneous Med (Zagreb). 2012;22(3):276-282.
infections due to nontuberculous mycobacteria: histopatho- 16. Elston DM, Stratman EJ, Miller SJ. Skin biopsy. Biopsy issues in
logical review of 28 cases. Comparative study between lesions specific diseases. J Am Acad Dermatol. 2016;74(1):1-16.
observed in immunosuppressed patients and normal hosts. 17. Gupta E, Bhalla P, Khurana N, Singh T. Histopathology for the
J Cutan Pathol. 2000;27:124-129. diagnosis of infectious diseases. Indian J Med Microbiol. 2009;
8. Gonzalez Santiago TM, Pritt B, Gibson LE. Diagnosis of deep 27(2):100-106.
cutaneous fungal infections: correlation between skin tissue 18. Chandler FW. Infectious disease pathology: morphologic and
culture and histopathology. J Am Acad Dermatol. 2014;71(2): molecular approaches to diagnosis. J Histotechnol. 1995;18:
293-301. 183-186.