Diagnostic values of KOH examination, histological
examination, and culture for onychomycosis: a latent class
analysis
Bilge Fettahlıog˘lu Karaman
1,
MD
, Arbil Ac
ßıkalın
2,
MD
, _Ilker €
Unal
3,
PhD
, and
Varol L. Aksungur
1,
MD
1Department of Dermatology, Faculty of
Medicine, Cßukurova University, Adana, Turkey,2Department of Pathology, Faculty
of Medicine, Cßukurova University, Adana, Turkey, and3Department of Biostatistics, Faculty of Medicine, Cß ukurova University, Adana, Turkey
Correspondence
Bilge Fettahlıog˘lu Karaman,MD Department of Dermatology, Faculty of Medicine
Cß ukurova University 01330, Adana Turkey
E-mail: bilgef@hotmail.com.tr
Funding: This study was supported by Research Fund of Cukurova University Project (TSA-2015-5032).
Conflict of interest: None. doi: 10.1111/ijd.14255
Abstract
Background In the absence of a real gold standard, comparative studies are still done on diagnostic methods for onychomycosis. There are only a few attempts using latent class analysis to determine the value of polymerase chain reaction in comparison to conventional methods. We aimed to determine the value of histological examination in such a way for the diagnosis of onychomycosis.
Methods Potassium hydroxide mount (KOH), culture and histological examination with periodic acid-Schiff (PAS), and Gomori’s methenamine silver (GMS) stains were done in 106 patients having clinically suspected toenail onychomycosis.
Results KOH was positive in 74% of the patients; culture in 14%; PAS in 30%; and GMS in 66%. According to the results of the latent class analysis, culture and PAS were highly specific but poorly sensitive; KOH, highly sensitive but poorly specific; and GMS, both highly sensitive and specific.
Conclusions Based on these results, we have proposed KOH as a screening test and GMS as a confirmatory test for the diagnosis of onychomycosis in our own practice. However, since positivity rates of different diagnostic methods vary widely in different centers, it is more suitable that every center should determine their own diagnostic strategy by evaluating their own results with latent class analysis.
Introduction
Nowadays, the main diagnostic methods for onychomycosis are direct microscopic examination, histological examination, cul-ture, and polymerase chain reaction (PCR).1,2 They differ in
rapidness, ability to identify fungal species, and cost. Culture is the slowest method. Direct microscopic examination and histo-logical examination are rapid but cannot identify fungal species. PCR is both rapid and capable of identifying the species. How-ever, it has a relatively high cost.
Even in the most recent years, comparative studies have still been done to evaluate the value of these methods in the diag-nosis of onychomycosis.3–6However, there is no real gold stan-dard to calculate their sensitivities and specificities. So, latent class analysis (LCA) has been used to determine the value of PCR in the diagnosis of onychomycosis in comparison to direct microscopic examination and culture.7,8
Since the vast majority of toenail onychomycosis are due to dermatophytes9 and all of the dermatophyte species respond
similarly to the available major antifungals,10 identification of
fungal species can be neglected at the beginning of the treat-ment in a busy daily clinical practice. So, histological examina-tion may be preferred instead of PCR. Then, it is necessary to determine the value of histological examination in comparison to direct microscopic examination and culture, again using LCA.
Materials and methods
Direct microscopic examination, histological examination, and culture were performed on samples taken from 106 consecutive patients having changes clinically suggestive of onychomycosis, namely discoloration, dystrophy, thickening, subungual hyperkeratosis, and onycholysis, in great toenails. Fifty-nine patients (55.7%) were female, and 47 (44.3%), male. Their mean age was 52.7 years (SD 14.3, range 17–86 years). Duration of the toenail changes varied from 1 month to 40 years.
Samples were taken only from one involved great toenail. If both great toenails were involved, the more severely involved one was chosen for sampling. The diseased area of the great
therefrom. Scrapings were used for direct microscopic examination and culture, and fragments for histological examination. So, we used one part or another for each test, namely distal parts of the diseased nail for direct microscopic examination and culture and its proximal parts for histological examination.
Direct microscopic examination was done by wet mount preparation with 20% KOH. Only if hyphae, namely thread-like structures brighter than the background, having two parallel borders and crossing the edges of corneocytes, were seen, KOH mount was designated as positive (Fig. 1). Greenish tint on focusing up and down, branching and septating were not accepted as mandatory but as supportive for hyphae.
Culture was done using Sabouraud’s glucose agar (Merck, Darmstadt, Germany) on plates containing chloramphenicol (0.1 mg/ml) and gentamicin (0.05 mg/ml) with and without cycloheximide (0.1 mg/ml). The plates were incubated for
2–4 weeks at 28 °C and evaluated weekly. If a growth typical for filamentous fungi was observed, the colonies were examined with lactophenol cotton blue stain under the microscope. Non-dermatophytes, dermatophytes, and also Candida species were recorded according to microscopic and macroscopic features of the colonies. Further identification procedures were performed only for dermatophyte strains since the repetitive samples could not be obtained from the patients to confirm the non-dermatophyte infections. Dermatophytes were inoculated on urea broth and incubated at 28°C for a week in order to test urease activity. Identification of the dermatophytes was done based on growth rate, morphological features of colonies, characteristics of microconidia,
macroconidia, and hyphae on microscopic examination, and urease activity.
For histological examination, nail fragments were fixed for 4 hours in 10% buffered formalin and were then soaked in 10% nitric acid solution until they were softened. A fully automated tissue tracking and paraffin-embedding process was done. Three-micrometer slices were stained with periodic acid–Schiff (PAS) and Gomori’s methenamine silver (GMS) stains. On light microscopy, the presence of purple-red dots or thread-like structures in PAS-stained sections and black-brown ones in GMS-stained sections were accepted to be positive for the presence of fungi (Fig. 2).
In statistical evaluation, an LCA was performed using the poLCA library11in the R statistical package (version 3.3.3, The R Foundation for Statistical Computing). Sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy of all four diagnostic methods used in this study, namely KOH, culture, PAS, and GMS, were calculated with the results of the LCA. Comparison of frequencies was analyzed by chi-square test.
(a)
(b)
Figure 1 On KOH examination, hyphae are seen as branching thread-like structures which are brighter than the background and slightly greenish (9400). (a) Their borders are darker, and their insides are brighter at a focusing level. (b) In contrast, their borders are brighter and their insides are darker at another focusing level
Figure 2 On histological examination with GMS, hyphae are seen as blackish dots or thread-like structures on a pale and mostly green background consisting of corneocyte layers of the nail plate (9400)
Written informed consent was obtained from all patients. This study was approved by the Ethics Committee of the Medical Faculty of Cßukurova University and supported by Research Fund of Cukurova University, Adana, Turkey. Project Number: TSA-2015-5032.
Results
The clinical appearance was consistent with the following sub-types of onychomycosis, “distal and lateral subungual” in 72 patients (67.9%), “total dystrophic” in 33 patients (31.1%), and “proximal subungual” in 1 patient (0.9%). The frequency of posi-tivity was 73.6% for KOH, 14.2% for culture, 30.2% for PAS, and 66.0% for GMS. Trichophyton rubrum was isolated in 12 patients, non-dermatophyte molds in 2 patients, and Candida in 1 patient. In all except one, patients with growth of Trichophyton rubrum on culture, at least one of the other tests was positive. Both patients with growth of nondermatophyte molds were also GMS-positive. Moreover, one of them was also KOH-positive. However, in the patient with growth of Candida on culture, all other tests were negative. In all PAS-positive patients, GMS was also positive. In all of these four diagnostic methods, fungi were found more frequently in patients clinically consistent with total dystrophic onychomycosis than in patients clinically consis-tent with distal and lateral subungual onychomycosis. However, the difference was statistically significant only in GMS (81.8% vs. 59.7%, Pffi 0.02).
Because we had four results from the diagnostic methods, each being either positive or negative, this system gave a possi-bility of 16 (24) combinations (Table 1). All of the four tests were negative in 13 patients (12.3%) and were positive in 4 patients (3.8%). Solitary positivity was found for KOH in 20 patients (18.9%), for culture in two patients (1.9%), and for GMS in eight
patients (7.5%) but not for PAS. The most common combina-tions were combined positivities for KOH and histological exami-nations. Combined positivity was encountered for KOH and both histological examinations in 24 patients (22.6%) and for KOH and only GMS in 23 patients (21.7%) but not for KOH and PAS.
The results of the LCA was given in Table 1. Sensitivity, specificity, positive and negative predictive values, and diagnos-tic accuracy of the four diagnosdiagnos-tic methods, which were calcu-lated with the results of the LCA, are listed in Table 2. The lowest sensitivity (19%) and negative predictive value (45%) were observed in culture and the highest ones (100% and 100%) in GMS. The lowest specificity (52%) and positive pre-dictive value (73%) were observed in KOH, and the highest ones (100 and 100%) in PAS. GMS had the highest diagnostic accuracy (93%) and culture, the lowest (50%). The intermediate diagnostic accuracies were seen in KOH and PAS (76 and 72%, respectively).
Discussion
A recently published meta-analysis has evaluated the utility of direct microscopic examination, culture, and histological exami-nation for the diagnosis of onychomycosis.12In the included
stud-ies of this meta-analysis13–22 and similar studies,5,6,23 the
positivity rates of these diagnostic methods have shown wide ranges (Table 3). This rate was ranged from 23 to 77% for KOH, from 10 to 68% for culture, and from 38 to 80% for PAS. We found the frequency of positivity to be 74% for KOH, and this fig-ure was within the range of the previous reports. However, our figure for culture (14%) was just above the lower limit of the range, and for PAS (30%), under the lower limit. On the other hand, we found GMS to be superior to PAS as in the study by
Table 1 Frequencies of combinations derived from either positivity or negativity of the four diagnostic methods for onycomycosis and the results of the LCA
Combination No. KOH Culture PAS GMS Frequency (No. of patients) Results of LCA
1 13 Negative 2 + 8 Negative 3 + 0 4 + + 3 Positive 5 + 2 Negative 6 + + 1 Positive 7 + + 0 8 + + + 1 Positive 9 + 20 Negative 10 + + 23 Positive 11 + + 0 12 + + + 24 Positive 13 + + 1 Negative 14 + + + 6 Positive 15 + + + 0 16 + + + + 4 Positive
D’Hue et al.24Its positivity rate (66%) was within the range for histological examination obtained from the previous reports.
If one puts the diagnostic methods in an increasing order according to positivity rates for each of the aforementioned reports, the most common order will be “culture, direct micro-scopy, and histology.” The order in our study, namely “culture, histology, and direct microscopy”, will be seen to be the second most common order. Briefly, culture had the lowest positivity rates in the majority of these reports comparing it to direct microscopy and histology for the diagnosis of onychomycosis. In spite of this fact, our very low positivity rate for culture is an issue requiring a further explanation.
One possible explanation was that we first scraped the dis-eased nail and then clipped a fragment from it. So, the scraping
was taken from the distal part and the fragment from the proxi-mal part. Culture was done with the scraping, in other words a specimen taken from the distal part. In a study, nail samples were obtained from proximal, medial, and distal parts of the nail in each of 194 patients by two methods, namely curettage and drilling.25Positivity rates both for KOH and culture increased as
the location of the sample was more proximal. For example, if the specimen was taken from the distal part by curettage, cul-ture was positive in 11.4% of patients, whereas if the specimen was taken from the proximal part, culture was positive in 61.8% of patients. The rate obtained from the distal part was very close to the positivity rate for culture (14.2%) in our study. How-ever, our rate for KOH (73.6%), which was also done with the specimen taken from the distal part, was higher even than the
Table 2 Sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy of the four diagnostic methods for onychomycosis, which were calculated with the results of the LCA
Indices KOH Culture PAS GMS
Sensitivity % (CI 0.95) 92 (82–97) 19 (10–31) 52 (39–65) 100 (94–100)
Specificity % (CI 0.95) 52 (37–68) 93 (81–99) 100 (92–100) 82 (67–92)
Positive predictive value % (CI 0.95) 73 (62–83) 80 (52–96) 100 (89–100) 89 (79–95)
Negative predictive value % (CI 0.95) 82 (63–94) 45 (35–56) 60 (47–71) 100 (90–100)
Diagnostic accuracy % (CI 0.95) 76 (66–83) 50 (40–60) 72 (62–80) 93 (86–97)
Table 3 Positivity rates of KOH, culture, and histological examination with PAS in studies comparing these methods in subjects with clinical suspicion or clinical signs of onychomycosis
Author Year N
Rate of positivity (%)
Increasing order by these rates KOH Culture PAS Dtm Sab Myc Lawry et al.13 2000 63 40a 24 63 C< D < P Borkowski et al.14 2001 50 32 10 20 72 C< D < P Gianni et al.23 2001 172 59 52 55 C< P < D Weinberg et al.15 2003 105 66 48c 74 C< D < P Karimzadegan-Nia et al.16 2007 96 38 26 40 C< D < P Hsiao et al.17 2007 88 77 59 72 C< P < D Shenoy et al.18 2008 101 53 35 75 C< D < P Alkhayat et al.19 2009 141 23 41c 38 D< P < C Jung et al.20 2015 93 41 22d 65 C< D < P Jeelani et al.21 2015 216 64b 51 68 76 D< C < P Hajar et al.22 2015 192 59a 18 74 C< D < P Lavorato et al.5 2017 212 58b 34c 40 C< P < D Sipra et al.6 2017 300 60 53 80 C< D < P Our findings 2017 106 74 14 30 C< P < D
Sab, sabouraud agar; Dtm, dermatophyte test medium; Myc, mycosel agar; C, culture; D, direct microscopic examination; P, histological examination with PAS.
a
In this study, KOH was done with chlorazol black E.
b
In this study, KOH was done with dimethyl sulfoxide.
c
In this study, culture was done not only in Sabouraud but also in other agars. However, only one rate of positivity was given for culture results.
d
rate (63.4%) obtained from the proximal part by curettage in the aforementioned study.
Another possible explanation for our very low positivity rate in culture may emerge from the wide ranges of the positivity rates in the diagnostic methods for onychomycosis. It is most likely that there are different levels of skill for different tests in differ-ent cdiffer-enters. For example, a cdiffer-enter may have a greater skill for culture than for KOH, so they get a higher positivity rate for cul-ture, as in the study of Alkhayat et al. (23% for KOH and 41% for culture).19 On the other hand, another center may have a
greater skill for direct microscopy, so they get a higher positivity rate for direct microscopy, as in the study of Hajar et al. (59% for direct microscopy and 18% for culture).22
Even though our very low positivity rate may be explained by the above arguments, culture is “the weak link in the chain”, in other words a limitation of the present study, since we omitted repetitive cultures to confirm the growth of nondermatophyte molds and omitted further tests to identify the species of Can-dida. On the other hand, two patients with growth of nonder-matophyte molds on culture also showed fungi within the nail plate on histological examination. So, the possibility that these growths were due to laboratory or environmental contaminants was low. Even though we assume that these growths are due to causative agents, our very low positivity rates do not allow us to make a conclusion about frequencies of causative agents in toenail onychomycosis. However, we found dermatophytes, namely Trichophyton rubrum, to be the most common causative agents, as in a previous study done in our region. _Ilkit analyzed results of mycological tests of nail specimens collected in the period between 1998 and 2002.26He isolated dermatophytes in 68 (79%) out of 86 culture-positive toenail specimens. On the other hand, again from our country but from another region, Cengiz et al. recently reported that dermatophytes were isolated in 124 (55%) out of 225 culture-positive fingernail or toenail specimens.27So, even in the same country, frequencies of
cau-sative agents in onychomycosis may show regional differences. In most comparative studies on the diagnostic methods for onychomycosis, the “gold standard” has been defined to be at least one positivity in the methods studied. So, it is impossible to calculate some diagnostic indices, especially specificity and positive predictive value. On the other hand, in selection of screening and confirmatory diagnostic tests, it is important to know not only their sensitivities but also their specificities. So, in the absence of a real gold standard, it is more reasonable to use some statistical methods such as LCA, which gives a possi-bility to calculate not only sensitivities but also specificities. In two studies, PCR was compared to direct microscopic examina-tion and culture with LCA, and both its sensitivity and specificity were found to be over 80%.7,8
In the present study, according to the results of the LCA, cul-ture and PAS had a low sensitivity but a high specificity. In con-trast, KOH had a high sensitivity but a low specificity. Finally, GMS had high values for both of these indices. So, we can use
KOH as a screening test in patients having changes clinically suggestive of onychomycosis both because of its high sensitiv-ity and because of its being an office-based test. However, we need another test for confirmation because of its low specificity. At first glance, all of the rest are candidates for the confirmatory test because of their high specificities. However, since it is preferable that a confirmatory test should also have a high sen-sitivity, the most reasonable confirmatory test is GMS according to our present results.
In the article of a comparative study, Jung et al. emphasized that positivity rates of KOH and culture may vary depending on the skill of the physician, as we have done above.20In this study, they compared KOH, culture, and PAS with each other for the diagnosis of onychomycosis and calculated the sensitivity for each test using a “gold standard” defined to be at least one positivity. Then, a diagnostic algorithm was proposed taking into consideration the calculated sensitivities in addition to cost-effec-tiveness and time requirements for each test. Their algorithm is fairly different from our proposed strategy in the preceding para-graph. Although they did not use the LCA and so calculated the sensitivities in a different way from us, it was most likely that the difference between their reasoning and our reasoning was derived from the difference between their and our positivity rates for each test, so from the difference between their and our skills.
Because of the wide ranges of the positivity rate for each diagnostic method, which are most likely resulting from different levels of skill for different tests in different centers or resulting from different manners for sampling, transporting, and process-ing of clinical specimens in different centers, it seems not rational to purpose a global approach for the diagnosis of ony-chomycosis from the results of a study and even from the pooled results of similar studies. Instead of a global approach, it is more appropriate that every center should compare positivity rates of their available tests preferably with LCA in order to develop their own algorithm for the diagnosis of onychomycosis. On the other hand, attempts should be done to standardize sampling, transporting, and processing of clinical specimens so that comparable results will be obtained from studies done in different centers.
Acknowledgments
The authors thank Professor M. Macit Ilkit and Hazal Boral (Cß ukurova University, Department of Microbiology, Division of Mycology) for their critical laboratory support, help, and exper-tise. We also thank G€okcen Sahin, senior technician in the Cßukurova University, Department of Dermatology, for providing technical support in the direct microscopic examination.
References
1 Ghannoum M, Mukherjee P, Isham N, et al. Examining the importance of laboratory and diagnostic testing when treating
and diagnosing onychomycosis. Int J Dermatol 2018; 57: 131– 138.
2 Rios-Yuil JM. Onychomycosis laboratory diagnosis: review. Curr Fungal Infect Rep 2017; 11: 92–97.
3 Spiliopoulou A, Bartzavali C, Jelastopulu E, et al. Evaluation of a commercial PCR test for the diagnosis of dermatophyte nail infections. J Med Microbiol 2015; 64: 25–31.
4 Hafirassou AZ, Valero C, Gassem N, et al. Usefulness of techniques based on real time PCR for the identification of onychomycosis-causing species. Mycoses 2017; 60: 638–644. 5 Lavorato FG, Guimar~aes DA, Premazzi MG, et al. Performance
of mycology and histopathology tests for the diagnosis of toenail onychomycosis due to filamentous fungi: dermatophyte and non-dermatophyte moulds. Mycoses 2017; 60: 587–593.
6 Sipra MWK, Akhtar ZM, Jillani KS. Comparison of microscopy, culture and histopathology in the diagnosis of onychomycosis. PJMHS 2017; 11: 57–60.
7 Gong J, Ran M, Wang X, et al. Development and evaluation of a novel real-time PCR for pan-dermatophyte detection in nail specimens. Mycopathologia 2016; 181: 51–57.
8 Petinataud D, Berger S, Ferdynus C, et al. Optimising the diagnostic strategy for onychomycosis from sample collection to fungal identification evaluation of a diagnostic kit for real-time PCR. Mycoses 2016; 59: 304–311.
9 Scher RK, Rich P, Pariser D, et al. The epidemiology, etiology, and pathophysiology of onychomycosis. Semin Cutan Med Surg 2013; 32: S2–S4.
10 Hay RJ, Ashbee HR. Fungal infections. In: Griffiths CEM, Barker J, Bleiker T, Chalmers R, Creamer D, eds. Rook’s Textbook of Dermatology, 9th edn. New Delhi: Wiley-Blackwell, 2016: 32.1–32.96.
11 Linzer DA, Lewis JB. poLCA: an R package for polytomous variable latent class analysis. J Stat Soft 2011; 42: 1–29. 12 Velasquez-Agudelo V, Cardona-Arias JA. Meta-analysis of the
utility of culture, biopsy, and direct KOH examination for the diagnosis of onychomycosis. BMC Infect Dis 2017; 17: 166. 13 Lawry MA, Haneke E, Strobeck K, et al. Methods for diagnosing
onychomycosis: a comparative study and review of the literature. Arch Dermatol 2000; 136: 1112–1116.
14 Borkowski P, Williams M, Holewinski J, et al. Onychomycosis: an analysis of 50 cases and a comparison of diagnostic techniques. J Am Podiatr Med Assoc 2001; 91: 351–355.
15 Weinberg JM, Koestenblatt EK, Tutrone WD, et al. Comparison of diagnostic methods in the evaluation of onychomycosis. J Am Acad Dermatol 2003; 49: 193–197.
16 Karimzadegan-Nia M, Mir-Amin-Mohammadi A, Bouzari N, et al. Comparison of direct smear, culture and histology for the diagnosis of onychomycosis. Australas J Dermatol 2007; 48: 18–21.
17 Hsiao YP, Lin HS, Wu TW, et al. A comparative study of KOH test, PAS staining and fungal culture in diagnosis of
onychomycosis in Taiwan. J Dermatol Sci 2007; 45: 138–140. 18 Shenoy MM, Teerthanath S, Karnaker VK, et al. Comparison of
potassium hydroxide mount and mycological culture with histopathologic examination using periodic acid-Schiff staining of the nail clippings in the diagnosis of onychomycosis. Indian J Dermatol Venereol Leprol 2008; 74: 226–229.
19 Alkhayat H, Al-Sulaili N, O’Brien E, et al. The PAS stain for routine diagnosis of onychomycosis. Bahrain Med Bull 2009; 31: 1–8.
20 Jung MY, Shim JH, Lee JH, et al. Comparison of diagnostic methods for onychomycosis, and proposal of a diagnostic algorithm. Clin Exp Dermatol 2015; 40: 479–484. 21 Jeelani S, Ahmed QM, Lanker AM, et al. Histopathological
examination of nail clippings using PAS staining (HPE-PAS): gold standard in diagnosis of onychomycosis. Mycoses 2015; 58: 27–32.
22 Hajar T, Fernandez-Martinez R, Moreno-Coutino G, et al. Modified PAS stain: a new diagnostic method for onychomycosis. Rev Iberoam Micol 2016; 33: 34–37. 23 Gianni C, Morelli V, Cerri A, et al. Usefulness of histological
examination for the diagnosis of onychomycosis. Dermatology 2001; 202: 283–288.
24 D’Hue Z, Perkins SM, Billings SD. GMS is superior to PAS for diagnosis of onychomycosis. J Cutan Pathol 2008; 35: 745–747. 25 Shemer A, Trau H, Davidovici B, et al. Collection of fungi
samples from nails: comparative study of curettage and drilling techniques. J Eur Acad Dermatol Venereol 2008; 22: 182–185. 26 Ilkit M. Onychomycosis in Adana, Turkey: a 5-year study. Int J
Dermatol 2005; 44: 851–854.
27 Cengiz FP, Cevirgen BC, Emiroglu N, et al. Etiology of onychomycosis in patients, Turkey. J Am Podiatr Med Assoc 2018; 108: 253–256.
