Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
305
Original Article
SMJ
Chayada Chaiyabutr, M.D., Nuttagarn Jantanapornchai, M.D., Chalermkwan Apinuntham, M.D., Charussri
Leeyaphan, M.D., Sukhum Jiamton, M.D., Ph.D.
Department of Dermatology, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok 10700, ailand.
Comparison of the Sensitivity and Specicity of
Tzanck Smear and Immunouorescence Assay for
the Diagnosis of Cutaneous Herpes Simplex Virus
and Varicella Zoster Virus Infections in a Real-life
Clinical Setting
ABSTRACT
Objective: This research aims to compare (1) the sensitivity and specificity of Tzanck smear and indirect
immunouorescence assay (IFA) which detect viral antigen for the diagnosis of cutaneous herpes simplex virus
(HSV) and varicella zoster virus (VZV) infections; and (2) the detection rates of the tests among various patient
groups and lesion morphologies.
Materials and Methods: is retrospective study reviewed 440 and 172 samples from patients with clinically
suspicious cutaneous HSV and VZV infections, who underwent both Tzanck smear and IFA, respectively. e gold
standard of the study was dened by showing agreement of diagnostic codes between initial and subsequent visits.
Results: For HSV infections, the respective sensitivity and specicity of Tzanck smear were 32.8% and 96.6%
whereas those of IFA were 60.7% and 100%. As to VZV infections, the sensitivity and specicity of Tzanck smear
were 54.3% and 97.8%, respectively, while the corresponding values of IFA were 71.7% and 100%. According to
disease characteristics and lesion morphologies, IFA provided substantially higher ability to detect HSV than the
Tzanck smear, especially in patients with immunosuppressed conditions. Tzanck smear and IFA demonstrated no
statistically signicant dierence for early-onset VZV infections (≤ 3 days).
Conclusion: e Tzanck smear and IFA had higher sensitivities for detecting VZV than HSV infections. IFA testing
is recommended in patients with immunosuppressed conditions who present with suspected cutaneous HSV
infection. Despite the overall sensitivity and specicity of IFA being greater than those of Tzanck smear especially
in HSV infections, the latter test is comparable option for early-onset VZV infections.
Keywords: Herpes simplex virus; varicella zoster virus; Tzanck smear; immunouorescence (Siriraj Med J 2021;
73: 305-311)
Corresponding author: Sukhum Jiamton
E-mail: sukhum.jiamton@gmail.com
Received 25 September 2020 Revised 28 December 2020 Accepted 30 December 2020
ORCID ID: http://orcid.org/0000-0003-1068-1586
http://dx.doi.org/10.33192/Smj.2021.40
INTRODUCTION
Herpes simplex virus (HSV) and varicella zoster virus
(VZV) are large, enveloped DNA viruses belonging to the
Herpesviridae family.
1
Although cutaneous infections of
HSV and VZV are mainly diagnosed by history-taking
and clinical characteristics, laboratory examinations are
sometimes needed for a denite diagnosis.
2
Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
306
Multiple laboratory options currently available to
diagnose HSV and VZV infections can be categorized into
four groups: (1) morphological tests, such as the Tzanck
smear and tissue histopathology; (2) immunomorphological
tests, like immunouorescence and immunoperoxidase
staining; (3) serological methods, for instance, enzyme-
linked immunosorbent assay and immunoglobulin M/G
titer; and (4) virological testing, for example, viral culture
and viral polymerase chain reaction. A viral culture was
long considered the gold-standard diagnostic test before
the advent of polymerase chain reaction testing.
1,3
HSV and VZV are more commonly observed as
cutaneous infections rather than as infections of internal
organs.
1,4
Moreover, their cutaneous symptoms are usually
not severe and can be self-limited. Ideally, the chosen
diagnostic test for these infections should be easy to
perform, give a rapid result, and be inexpensive. In
outpatient dermatological settings, the Tzanck smear
and immunouorescence staining are therefore the most
frequently ordered tests at our clinic.
Previous research has found that the sensitivity of
Tzanck smear ranges from 34% to 78% in detecting HSV,
and from 26% to 64% in detecting VZV.
5
However, with
a procient technician, the sensitivity of the test may
rise to 80% and its specicity to 90%.
6-8
Even though
Tzanck smear is currently considered obsolete in many
countries
1
, it still has an important role in developing
countries. ere, the newer testing methodologies are
not only oen deemed to be too expensive, but also
have the drawbacks of slower turnaround times and,
sometimes, a lack of availability.
In terms of immunouorescence testing, previous
studies revealed that the sensitivity and specicity of
immunouorescence staining was greater than those of
Tzanck smear, particularly in the case of VZV infections.
e sensitivity of immunouorescence staining in detecting
cutaneous HSV infections was found to be around 50%
- 100% compared with the viral culture technique, but
its sensitivity in detecting VZV infections exceeded
that of the viral culture. Moreover, the specicity of
immunouorescence staining was nearly 100%, and it
was able to discriminate between the HSV1/2 and VZV
pathogens.
2,8
Earlier studies of the sensitivity and specicity of
the Tzanck smear and immunouorescence testing were
usually performed in a small number of patients, and
compared with those of the viral culture technique as the
gold standard diagnostic modality.
9
It is also noteworthy
that few details of the infected patients or the clinical
morphologies of their lesions were reported.
2,5,9
us,
the objective of the current research was twofold. e
rst aim was to compare the sensitivity and specicity
of the Tzanck smear and immunouorescence assay for
the diagnosis of cutaneous HSV and VZV infections in a
larger population and in a real-life setting. e secondary
aim of this study was to compare the detection rates of
two tests among various subgroups of patients, durations,
and clinical morphologies of lesions.
MATERIALS AND METHODS
is retrospective research was approved by the
Siriraj Institutional Review Board. (Si 333/2020) e
study reviewed the samples taken from patients with
clinically suspicious cutaneous HSV (ICD-10 B00) and
VZV (ICD-10 B01-B02) infections. During 2012-2019,
the samples had initially been collected from the Infection
Control Clinic of the Department of Dermatology, Faculty
of Medicine Siriraj Hospital, Mahidol University.
All eligible patients had to undergo both a Tzanck
smear and immunofluorescence testing for HSV or
VZV. For each patient, demographic data, onset of
lesions, morphology of the lesions, suspicious diagnosis,
and comorbidities had been collected at their rst visit.
Tzanck-smear and IFA specimens that were reported as
being inadequate for diagnosis were excluded from the
study. e sensitivity and specicity of the tests were
subsequently analyzed only in clinically conrmed cases.
For the purposes of this study, the reference standard
for clinically conrmed diagnosis was an agreement
of diagnostic codes between the rst and subsequent
follow-up visits (determined by a dermatologist).
e Tzanck smear was performed by scraping the
base of lesions with a blunt scalpel blade and spreading
the sample as a thin layer on microscope slides. e slides
were then xed with 100% methyl alcohol for 10 minutes
and stained with eosin solution for 20 seconds. Aer being
rinsed with distilled water, the slides were stained with
3% methylene blue for 60 seconds, rinsed with distilled
water, and allowed to dry. e slides were subsequently
examined under a light microscope. A positive Tzanck
smear was dened as the presence of herpetic cytopathic
eects, such as the presence of multinucleated giant
cells. roughout the 7-year study period, the Tzanck
tests were performed by the same, procient technician,
which obviated inter-rater variability. e typical test
turnaround time was 15 minutes.
As to the immunouorescence staining, our hospital
used the technique of an indirect immunouorescence
assay (IFA) with commercial reagent kit containing HSV
type 1, 2 antibodies (Bio-Rad Laboratories) and VZV
monoclonal antibodies (Merck, Ltd). Specimens scraped
from the base of the lesions were xed in acetone for 15
Chaiyabutr et al.
Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
307
Original Article
SMJ
minutes before adding a primary antibody. e smear
was then incubated at 37 degrees Celsius for 30 minutes
and rinsed with phosphate-buered saline; pathogen-
specic uorescein-tagged secondary antibodies were
subsequently added, and the mixture was incubated
at 37 degrees Celsius for 30 minutes. e smear was
examined with an epiuorescence microscope by virology
technicians. e test turnaround time was 3 days.
All statistical analyses were undertaken using SPSS
Statistics for Windows, version 18.0 (SPSS Inc., Chicago,
IL, USA). Categorical variables were presented as numbers
or numbers with percentages, while continuous variables
were shown as means with standard deviations. e
sensitivities and specicities of the two tests were calculated.
For the correlation between the tests, Cohen’s Kappa
coecient(
κ
) was reported. A p-value of less than 0.05
was considered to indicate statistical signicance.
RESULTS
A total of 440 and 172 specimens from patients with
clinically suspicious cutaneous HSV and VZV infections,
respectively, were reviewed. e demographic data of the
included patients are detailed in Table 1. e mean age
of the participants was around 50. e majority of them
were tested more than 3 days aer onset of the lesions.
For cutaneous HSV infections, the main characteristic
of the tested lesions was non-vesicle (66.2%) whereas
vesicle was the major type of tested lesions in cutaneous
VZV infections (70.1%).
TABLE 1. Demographic data and disease characteristics of the included patients.
HSV VZV
N = 440 N = 172
N (%) N (%)
Demographic
Sex, Female 239 (54.3) 103 (59.9)
Age (mean) ± SD 51.4 ± 18.4 56.1 ± 18.2
Underlying disease
Hypertension 101 (23.0) 36 (20.9)
Diabetes mellitus 49 (11.1) 20 (11.6)
Autoimmune disease 51 (11.6) 16 (9.3)
Cancer 51 (11.6) 32 (18.6)
HIV (n = 240; n = 62) 45 (18.7) 6 (9.7)
On immunosuppressive drugs 66 (15.0) 26 (15.1)
Disease characteristics
Onset ≤ 3 days (n = 420; n = 167) 165 (39.3) 75 (44.9)
Taken oral acyclovir before testing 29 (6.6) 21 (12.2)
Site
Mucosa 233 (53.0) 8 (4.7)
Skin 207 (47.0) 164 (95.3)
Morphology (n = 420; n = 164)
Vesicle 142 (33.8) 115 (70.1)
Non-vesicle 278 (66.2) 49 (29.9)
Erosion 113 (26.9) 6 (3.7)
Ulcer 105 (25.0) 3 (1.8)
Papule 26 (6.2) 19 (11.6)
Crust 12 (2.9) 12 (7.3)
Erythematous macule 11 (2.6) 8 (4.9)
Verrucous plaque 11 (2.6) 1 (0.6)
Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
308
Of 440 specimens, 229 (52%) had clinically conrmed
diagnosis of HSV, while 127 (73.8%) of 172 specimens
had clinically conrmed diagnosis of VZV. Table 2
compares sensitivities, specicities, positive predictive
values, and negative predictive values of the Tzanck smear
and IFA related to the clinically conrmed cases. For
HSV infections, the respective sensitivity and specicity
of Tzanck smear were 32.8% and 96.6% whereas those
of IFA were 60.7% and 100%. As to VZV infections,
the sensitivity and specicity of Tzanck smear were
54.3% and 97.8%, respectively, while the corresponding
values of IFA were 71.7% and 100%. e sensitivity and
specicity of IFA was substantially higher than those
of the Tzanck smear. In addition, the Tzanck smear
and IFA showed higher sensitivities in detecting VZV
infections than HSV infections. e Kappa agreements
between the Tzanck smear and IFA in detecting HSV
and VZV infections were moderate, with the values of
0.4 and 0.5, respectively.
A comparison was made on the sensitivity of the
Tzanck smear and IFA for cutaneous HSV and VZV
infections among various subgroups of patients, durations,
and clinical morphologies of lesions.
In cutaneous HSV infections (Table 3), it was found
that IFA yielded a greater sensitivity in detecting HSV
infections than the Tzanck smear in nearly all subgroups
of patients with statistical signicance. In terms of disease
onset, IFA showed the sensitivity around 60% in both
early-onset (≤ 3 days) and late-onset (> 3 days) HSV
infections whereas the percentage of HSV detection from
the Tzanck smear dropped from 45.2% in early-onset to
24.8% in late-onset HSV infections. Furthermore, the
sensitivity rate of Tzanck smear in patients taken oral
acyclovir before testing was very low (19%) while IFA in
these patients still yielded a sensitivity rate nearly 60%.
Clinical morphologies of the lesions also determined the
sensitivity rates of both tests. IFA also showed a high
sensitivity (around 60%) in detecting HSV in vesicle
and non-vesicle lesions. e detection rate of Tzanck
smear was only 45.7% in vesicle lesions and very low in
non-vesicle lesions (25.4%).
In cutaneous VZV infections (Table 4), even though
IFA yielded a greater sensitivity than Tzanck smear but
the magnitude of dierence was not much as in case of
cutaneous HSV infections. Interestingly, the Tzanck
smear was not statistically dierent from the IFA in some
situations such as early-onset (≤ 3 days) of infection,
non-vesicular lesions and patients who had a history
of taking oral acyclovir before testing.
DISCUSSION
e current research demonstrated that both the
Tzanck smear and IFA had a higher sensitivity in detecting
VZV infections than HSV infections. e comparison of
their sensitivities and specicities revealed that the IFA
was superior overall to the Tzanck smear, corresponding
with earlier ndings.
3
e higher sensitivity of IFA was
signicantly shown in nearly all situations of cutaneous
HSV infections.
However, in cutaneous VZV infections, the sensitivity
rate of Tzanck smear was not far dierent from IFA. Our
study showed that the Tzanck smear is still comparable
to the IFA in some VZV-infection situations, such as the
early-onset (≤ 3 days) of infection. is can be explained
by the Tzanck smear having a high ability to detect VZV
infections
5
, as well as by a shorter duration of disease
normally resulting in an increase in the sensitivity of
the Tzanck smear.
10
e type of lesions is also an important factor in
determining the sensitivity of the two tests. Prior research
TABLE 2. e sensitivities, specicities, positive predictive values (PPV), and negative predictive values (NPV) for
the cutaneous HSV and VZV infections.
HSV infection % Sensitivity % Specicity PPV* NPV**
Tzanck 32.8 98.6 96 57.5
IFA 60.7 100 100 70.1
VZV infection
Tzanck 54.3 97.8 98.6 43.1
IFA 71.7 100 100 55.6
Abbreviations: *PPV: positive predictive value, **NPV: negative predictive value
Chaiyabutr et al.
Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
309
Original Article
SMJ
TABLE 3. Comparison of the sensitivities of detection of the Tzanck smear and IFA for cutaneous HSV infections.
Positive Tzanck smears Positive IFAs P-value
N = 229 N = 229
N (%) N (%)
Demographics
Sex F 54/139 (38.8) 88/139 (63.3) < 0.001
M 21/90 (23.3) 51/90 (56.7) < 0.001
Age ≤ 60 years 46/145 (31.7) 87/145 (60.0) < 0.001
Age > 60 years 29/84 (34.5) 52/84 (61.9) < 0.001
Underlying disease
Hypertension 15/57 (26.3) 33/57 (57.9) 0.001
Diabetes mellitus 9/28 (32.1) 18/28 (64.3) 0.022
Autoimmune disease 6/25 (24.0) 17/25 (68.0) 0.003
Cancer 10/22 (45.5) 14/22 (63.6) 0.388
HIV 8/30 (26.7) 22/30 (73.3) 0.001
On immunosuppressive drugs 9/32 (28.1) 24/32 (75.0) < 0.001
Disease characteristics
Onset ≤ 3 days 42/93 (45.2) 59/93 (63.4) 0.003
Onset > 3 days 31/125 (24.8) 76/125 (60.8) < 0.001
Taken oral acyclovir before testing 4/21 (19.0) 12/21 (57.1) 0.008
Site
Mucosa 26/102 (25.5) 56/102 (54.9) < 0.001
Skin 49/127 (38.6) 83/127 (65.4) <0.001
Morphology
Vesicle 46/102 (45.1) 67/102 (65.7) 0.001
Non-vesicle 28/116 (24.1) 65/116 (56.0) <0.001
Ulcer 12/54 (22.2) 29/54 (53.7) <0.001
Erosion 8/41 (19.5) 24/41 (58.5) <0.001
Hypertrophic 4/9 (44.4) 6/9 (66.7) 0.625
Crust 2/6 (33.3) 2/6 (33.3) 1.000
Papule 2/4 (50.0) 2/4 (50.0) 1.000
Erythematous macule 0/2 (0.00) 2/2 (100) -
Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
310
TABLE 4. Comparison of the sensitivities of detection of the Tzanck smear and IFA for cutaneous VZV infections.
Positive Tzanck smears Positive IFAs P-value
N = 127 N = 127
N (%) N (%)
Demographics
Sex F 43/76 (56.6) 56/76 (73.7) 0.011
M 25/51 (49.0) 35/51 (68.0) 0.021
Age ≤ 60 years 40/69 (58.0) 50/69 (72.5) 0.031
Age > 60 years 28/58 (48.3) 41/58 (70.7) 0.007
Underlying disease
Hypertension 16/31 (51.6) 25/31 (80.6) 0.012
Diabetes mellitus 10/16 (62.5) 14/16 (87.5) 0.125
Autoimmune disease 7/12 (58.3) 11/12 (91.7) 0.125
Cancer 10/21 (47.6) 14/21 (66.7) 0.344
HIV 3/5 (60.0) 4/5 (80.0) 1.000
On immunosuppressive drug 15/19 (78.9) 15/19 (78.9) 1.000
Disease characteristics
Onset ≤ 3 days 40/58 (69.0) 47/58 (81.0) 0.092
Onset > 3 days 27/67 (40.3) 43/67 (64.2) 0.002
Taken oral acyclovir before testing 6/20 (30.0) 10/20 (50.0) 0.344
Site
Mucosa 1/3 (33.3) 3/3 (100) –
Skin 67/124 (54.0) 88/124 (71.0) 0.001
Morphology
Vesicle 55/95 (57.9) 75/95 (78.9) <0.001
Non-vesicle 8/26 (30.8) 13/26 (50.0) 0.180
Crust 4/10 (40.0) 7/10 (70.0) 0.250
Papule 2/9 (22.2) 3/9 (33.3) 1.000
Erosion 1/3 (33.3) 0/3 (00.0) -
Erythema 1/2 (50.0) 1/2 (50.0) 1.000
Ulcer 0/2 (00.0) 2/2 (100) -
Chaiyabutr et al.
Volume 73, No.5: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
311
Original Article
SMJ
has found that vesicles and blisters generally yield higher
sensitivities of detection than other types of lesions
with these two tests
2,10
; the present study had a similar
nding. However, our work determined that there was
no statistical dierence in the sensitivities of detection of
the Tzanck smear and IFA for non-vesicular lesions of
VZV. It is possible that the non-vesicular lesions which
were usually in the late stage of infection might have a low
number of virus and was therefore comparable dicult
for both tests to yield the positive result
1
, or the number
of specimens enrolled in the non-vesicular-VZV group
might not be enough to provide a statistically signicant
dierence.
Furthermore, in terms of underlying disease of the
patients, the sensitivity of IFA in cutaneous HSV infections
was prominently higher with statistical signicance
compared to Tzanck smear particularly in patients with
immunosuppressive conditions including HIV infection
and taking immunosuppressive agents. e detection
ability of HSV by Tzanck smear in these patients was
around 30% which was substantially lower than IFA (above
70%). IFA testing in suspected cutaneous HSV patients
with immunosuppressed conditions is recommended.
Whether the underlying disease would aect the yield of
Tzanck smear or IFA test in cutaneous VZV infections
was dicult to conclude. As the majority of underlying
diseases or comorbidity subgroups in cutaneous VZV
infections contained a small number of patients.
There are some limitations in this study. The
reference standard for confirmed cases used in this
study was a clinical diagnosis by dermatologists on two
separate occasions, rather than a standard laboratory
investigation like viral culture or polymerase chain reaction
testing. e explanation is that this was a retrospective
study conducted at a dermatology outpatient clinic in
a developing country and in a real-life clinical setting,
where dermatologists need to make prompt diagnosis
without the ready utilization of sophisticated laboratory
testing. For example, the use of viral culture tends to
be avoided because specimens need to be promptly
transported on ice to a laboratory, refrigerated-culture
media are required, and long turnaround times are
involved. Polymerase chain reaction testing, generally
recognized as the platinum standard for VZV and HSV
infections, has a higher sensitivity and specicity than
any other test. Nevertheless, its relatively high cost and
limited accessibility are problematic for developing
countries.
In addition, the number of patients with morphology
of vesicle were substantially higher in VZV (70.1%) than
HSV (33.8%). is might aect the overall sensitivity of both
tests and was another limitation of our study. However,
focusing in subgroup analysis based on morphology of
the lesions, Tzanck smear and IFA still yielded higher
sensitivity in VZV than HSV in either vesicle or non-
vesicle subgroup.
In conclusion, this study revealed the sensitivity and
specicity of the Tzanck smear and IFA which could be
used as a benchmark in a real-life setting. e tests had
a higher sensitivity in detecting VZV infections than
HSV infections. Even though IFA had an overall higher
sensitivity and specicity than the Tzanck smear, the
Tzanck smear is a comparable option to IFA for early-
onset VZV infections. is information is valuable,
especially in an outpatient dermatologic clinic, where
prompt diagnosis of HSV and VZV infections is required.
ACKNOWLEDGMENT
e authors thank Ms. Orawan Supapueng for her
assistance with the statistical analyses.
Conicts of interest: All authors have neither conicts
of interest nor nancial support to declare.
Funding: None
REFERENCES
1. Levin MJ, Weinberg A, Schmid DS. Herpes Simplex Virus and
Varicella-Zoster Virus. Microbiol Spectr, 2016.p.4.
2. Zirn JR, Tompkins SD, Huie C, Shea CR, Rapid detection
and distinction of cutaneous herpesvirus infections by direct
immunouorescence. J Am Acad Dermatol 1995;33: 724-8.
3. Frisch S, Guo AM, Diagnostic methods and management
strategies of herpes simplex and herpes zoster infections. Clin
Geriatr Med 2013;29:501-26.
4. Stuart-Harris C, e epidemiology and clinical presentation
of herpes virus infections. J Antimicrob Chemother 1983;12
Suppl B:1-8.
5. Nahass GT, Goldstein BA, Zhu WY, Sering U, Penneys NS,
Leonardi CL, Comparison of Tzanck smear, viral culture,
and DNA diagnostic methods in detection of herpes simplex
and varicella-zoster infection. JAMA 1992;268:2541-4.
6. Oranje AP, Folkers E, e Tzanck smear: old, but still of
inestimable value. Pediatr Dermatol 1988;5:127-9.
7. Grossman MC, Silvers DN, e Tzanck smear: can dermatologists
accurately interpret it? J Am Acad Dermatol 1992; 27:403-5.
8.
Fan F, Day S, Lu X, Tang YW, Laboratory diagnosis of HSV and
varicella zoster virus infections. Future Virology 2014;9:721-31.
9. Sadick NS, Swenson PD, Kaufman RL, Kaplan MH, Comparison
of detection of varicella-zoster virus by the Tzanck smear,
direct immunouorescence with a monoclonal antibody, and
virus isolation. J Am Acad Dermatol 1987;17:64-69.
10. Ozcan A, Senol M, Saglam H, Seyhan M, Durmaz R, Aktas E,
et al. Comparison of the Tzanck test and polymerase chain
reaction in the diagnosis of cutaneous herpes simplex and
varicella zoster virus infections. Int J Dermatol 2007;46:1177-9.
