An online quiz uncovers limitations of morphology in equivocal lung cytology

Authors


  • This study has been conducted according to the guidelines of the Ethical Committee of the University Hospital of Basel.

Abstract

BACKGROUND.

Equivocal atypia in respiratory cytology can be a diagnostic challenge. In such cases fluorescence in situ hybridization (FISH) may be used for the analysis of chromosomal aberrations and often allows a reliable distinction of benign and malignant cells.

METHODS.

An online picture gallery of 30 respiratory cytologic preparations comprising 23 specimens with equivocal cytology as well as 5 positive and 2 negative controls was prepared (www.unibas.ch/patho/lungenzyto/loesung/). The final diagnoses were confirmed by clinical follow-up or biopsy or both. Each of the illustrated cell groups was analyzed by multitarget FISH after PAP image capturing and automatic relocalization.

RESULTS.

The online questionnaire was completed by 137 cytomorphologists from all continents. The control cases were assessed accurately to a significantly higher percentage than the equivocal cases. In equivocal cases participants more often made false-positive than false-negative diagnoses. In 2 patients with benign conditions (tuberculosis and pulmonary capillaritis) the rate of false-positive answers was remarkably high (31.4% and 62.8% respectively). The result of the 20 best-performing participants for the 5 cases with the highest percentage of inaccurate answers was not better than if they had chosen their answer by chance.

CONCLUSIONS.

These data illustrate that single cells or cell clusters of a subgroup of equivocal lung cytology are a diagnostic challenge even for highly experienced morphologists. Internet-based tests are able to reveal limitations of cytomorphology. Cancer (Cancer Cytopathol) 2006; © 2006 American Cancer Society.

Equivocal cellular atypia is a common problem in both respiratory1–4 and nonrespiratory cytologies. A retrospective analysis of a 4-year period at our institution revealed that a diagnosis of equivocal atypia was made in 2.9% of all patients with respiratory cytology specimens, making it impossible for the cytologists to clearly classify those specimens as benign or malignant.5 In previous publications, we and others have demonstrated that multitarget fluorescence in situ hybridization (FISH) may be a useful diagnostic adjunct in lung cytology.5–9 In a study of 54 equivocal respiratory cytologic specimens FISH reached a sensitivity of 79% and a specificity of 100% for detection of malignancy.5 The aim of this survey was to explore to what extent equivocal lung cytology specimens can be correctly classified by cytomorphology alone depending on the expertise of the observer. For this purpose we collected 23 equivocal lung cytology cases and 7 lung cytology specimens with unequivocal diagnoses serving as controls. All cell groups shown in the quiz had previously been analyzed by multitarget FISH. For the quiz we selected cases that had been correctly classified by FISH as either malignant or benign. These 30 cases were assembled to an Internet-based quiz to compare the performance of international morphologists with interest in the field.

In addition, we wanted to test the acceptance of this internet-based quiz format for continuing education.

MATERIALS AND METHODS

Quiz Cases

Specimens were selected from the cytopathology files of the Institute for Pathology (University Hospital of Basel and the Cantonal Hospital of St. Gallen, Switzerland).

The 30 quiz cases included 13 bronchial secretions, 6 transbronchial fine-needle aspirates (FNAs), 5 bronchoalveolar lavages, 4 bronchial brushes, and 2 sputum specimens from 10 patients with a benign lung disorder and from 20 patients with a diagnosis of a carcinoma. The equivocal quiz cases were a subset of the 54 cytologic specimens of lung that had been analyzed in a previous study.5

The final diagnoses of patients with benign conditions were confirmed by a benign histology, and/or an uneventful clinical course, and/or a clinical diagnosis that could explain the atypical cytology. A final diagnosis of cancer was based on histology, autopsy, or unequivocal cytologic findings in additional specimens (Table 1). All atypical cells of carcinoma patients shown in the quiz had chromosomal aberrations detectable by FISH while reactive and benign cells showed a normal FISH result. The quiz included 23 equivocal and 7 control cases. A specimen was termed equivocal if at least 2 of 4 experienced cytopathologists (LB, PD, GF, or PS) had evaluated the original slides and considered the cytomorphologic findings as atypical and could not reach a firm diagnosis.

Table 1. Patients, Initial Clinical Diagnoses, Final Quiz Diagnoses, and Results of Participants
No.*Age/SexSpecimen typeInitial clinical diagnosisFinal diagnosis§Confirmation of diagnosisQuiz diagnosisAnswers
  • The column on the far right shows the percentage of correct answers. The 5 cases with the lowest percentage of correct results are indicated in bold print.

  • F indicates female; BAL, bronchoalveolar lavage; BS, bronchial secretion; SCLC, small-cell lung cancer; FNA, transbronchial fine-needle aspirate; NSCLC, non–small-cell lung cancer; M, male; BR, bronchial secretion; BB, bronchial brushing.

  • All specimens were direct smears.

  • *

    C indicates control cases with unequivocal diagnosis; E, equivocal diagnosis as defined in Materials and Methods section.

  • Participants comprised cytologists, cytotechnicians, surgical pathologists, and 4 pulmonologists. The answers “benign” and “favor benign” for benign lesions and the answers “carcinoma small cell” and “carcinoma nonsmall cell” and “favor malignant” for malignant lesions were counted as approximately correct answers.

  • The percentage of approximately correct answers is given in parentheses in this column.

  • §

    Numbers in parentheses (in this column) are months of follow-up for benign lesions.

1E83/FBALSuspicion of aspergillomaPneumonia with aspergilloma,  infiltrate resolvesClinicalBenign20.4 (59.1)
2E24/FBALInfiltratePneumonia (7)BiopsyBenign27.7 (64.2)
3C65/FBSSuspicion of lung cancerSmall-cell lung cancerBiopsySCLC70.1 (86.2)
4E77/FFNASuspicion of lung cancerAdenocarcinomaCytologyNSCLC79.6 (90.6)
5C65/FBALMetastasizing adenocarcinoma of the lungAdenocarcinomaCytologyNSCLC74.4 (92.0)
6E68/FFNASuspicion of lung cancerAdenocarcinomaBiopsyNSCLC9.5 (67.1)
7E58/MFNASolitary pulmonary noduleHamartomaBiopsyBenign8.8 (38.0)
8C79/FSputumNone indicatedAdenocarcinomaAutopsyNSCLC43.8 (65.7)
9E68/MSputumHemoptysisHeart failure, hemoptysis stops under  cardiac medication (1)ClinicalBenign19.7 (42.3)
10C64/FBRMultiple pulmonary nodulesUneventful courseBiopsyBenign86.9 (95.7)
11E40/MBALNSCLC after neoadjuvant  chemotherapyNon–small-cell lung cancerBiopsyNSCLC54.0 (75.9)
12C75/FBBSuspicion of lung cancerAdenocarcinomaCytologyNSCLC73.7 (97.8)
13E40/MFNASuspicion of lung cancerNon–small-cell lung cancerBiopsyNSCLC68.6 (94.2)
14C50/FFNASuspicion of lung cancerUneventful courseBiopsyBenign62.8 (75.9)
15E88/MBRPulmonary infiltrateUneventful course (13)ClinicalBenign43.1 (70.8)
16E60/MBRSolitary pulmonary noduleTuberculosis (3)ClinicalBenign11.7 (34.3)
17E71/MBRTumor of the left lower lobeSquamous-cell carcinomaCytologyNSCLC43.8 (73.0)
18C62/MBRSolitary pulmonary nodule right  upper lobeNon–small-cell lung cancerBiopsyNSCLC73.0 (98.5)
19E69/FBRSuspicion of lung cancerNon–small-cell lung cancerBiopsyNSCLC16.1 (38.7)
20E73/MBRLung tumorAdenocarcinomaBiopsyNSCLC56.9 (22.6)
21E61/MBRSuspicion of recurrent lung cancerSquamous-cell carcinomaBiopsyNSCLC29.9 (58.4)
22E76/MBRSuspicion of lung cancerSquamous-cell carcinomaBiopsyNSCLC19.0 (50.4)
23E83/MBBSuspicion of lung cancerSmall-cell lung cancerBiopsySCLC28.5 (63.5)
24E40/MBALHemoptysisCapillaritis (8)BiopsyBenign3.7 (11.7)
25E68/MBRLung abscess or tumorAdenocarcinomaBiopsyNSCLC32.8 (49.6)
26E62/FFNASuspicion of lung cancerSmall-cell lung cancerAutopsySCLC73.2 (43.8)
27E72/MBRPleural effusionAdenocarcinomaCytologyNSCLC24.1 (39.4)
28E73/FBBTumor of the right lungAdenocarcinomaBiopsyNSCLC26.3 (63.5)
29E75/FBRTumor of the right lower lobeBreast cancer metastasisBiopsyNSCLC20.4 (27.3)
30E65/MBBReticulo-nodular infiltrateInfiltrate resolves (39)ClinicalBenign31.4 (53.3)

Fluorescence In Situ Hybridization

All cell groups shown in the quiz had been analyzed by the commercially available multitarget FISH probe LA Vysion® (Abbott/Vysis; Downers Grove, IL) as described in detail previously.5 This FISH assay contains a mixture of fluorescent probes to the centromeric region of Chromosome 6, and to the 5p15, 8q24 (site of the MYC gene), and 7p12 (site of the EGFR gene) loci. A fraction of the probe mixtures was provided free of cost from Abbott/Vysis.

The Papanicolaou-stained atypical cell groups were photographed (AxioCam Color, Type 412-312), and the exact locations on the specimens were saved using an automated stage (Type 00-24-473-0000, Carl Zeiss AG, Oberkochen, Germany) on a Zeiss Axioplan 2 epifluorescence microscope (Zeiss, Jena, Germany), and a relocalization software (Mark&Find Module, Carl Zeiss Vision GmbH, Halbermoos, Germany). FISH was performed on Papanicolaou-stained preparations. Decoloration was achieved during denaturation and did not require a separate technical step. After relocalization, the hybridized atypical cells were selectively scored at 630× magnification. All cells of the atypical cell groups were evaluated. A locus or chromosome gain was indicated by a copy number of more than 2 signals of a probe. Specimens were defined as FISH-positive when there were at least 3 cells with multiple gains (gain of at least 2 chromosomal loci with 3 or more locus-specific signals). When only 1 of the locus-specific signals was affected, a minimum of 5 signals per nucleus in at least 3 cells were required for a positive result.

Quiz

The registration form and the online quiz cases were constructed by the questionnaire tool FlexiForm, which had been programmed at the Computing Department of the University of Basel. The quiz was subdivided into 7 sections. The first section comprised questions concerning basic personal data such as subspeciality training and experience of the participants. The following 6 sections contained 5 cases each.

Each of the 30 quiz cases comprised 1 to 5 images (4 on average), as well as information on specimen type and relevant clinical findings. All cell groups shown in the quiz had been analyzed by FISH using a relocalization software to make sure that the quiz diagnoses were based on the lesional cells. For each quiz case participants had to choose 1 of the following 6 answer categories: benign, atypical/unspecified, atypical/favor benign, atypical/favor malignant, carcinoma small cell, and carcinoma nonsmall cell.

In a first step, worldwide (cyto-) pathologists and cytotechnicians were invited to participate in a quiz on equivocal lung cytologic specimens. Study participants were recruited through personal communication, e-mail invitations, publication on the Web site of the Swiss Society of Pathology, and advertising during congresses or courses. Online registration was possible during the following 6 weeks. After the registration period, all enrolled individuals received an automatically generated e-mail with a personalized URL access to the quiz. Thanks to the personalized URL access, the participants had the option to interrupt the online test after any section and to return later to continue. The participants were given 2 months to submit their answers. After the deadline, the image gallery was no longer accessible online and the participants who had completed the quiz received a list with the correct answers for comparison with their own answers.

Participants were informed neither about the number of benign and malignant cases nor about the existence of control and atypical categories.

The answers of all participants were exported into an Excel file and were statistically evaluated anonymously by the Computing Department of the University of Basel.

Definition of Approximately Correct Answers

The answers “benign” and “favor benign” for benign lesions and the answers “carcinoma small cell” and “carcinoma nonsmall cell” and “favor malignant” for malignant lesions were counted as approximately correct answers.

Definition of Uncertain Answers

For each quiz case the percentage of participants opting for the answer categories “favor malignant”, “favor benign”, and “unspecified” was added up to assess the level of uncertainty.

Data Analysis

Contingency table analysis was used to study frequency comparisons of nominal categorized variables.

RESULTS

The quiz-cases, including the distribution of answers given by the participants, can be viewed online (http://pages.unibas.ch/patho/lungenzyto/loesung/).

Correct answers for each quiz-case are listed in Table 1.

Participants

The quiz was completed by 137 of the 263 individuals who had registered for participation. The 70 men and 67 women participants were from Switzerland (47), Italy (24), Germany (21), the Netherlands (8), Austria (6), and the USA (5). The remaining 26 were from 18 other countries. The participants indicated their profession as surgical pathologists with experience in cytopathology (64), cytopathologists (26), cytotechnologists/cytotechnicians (23), surgical pathologists without experience in cytopathology (10), residents in training for cytology (3) or pathology (7), and pulmonologists (4). They practiced at teaching hospitals/universities (67), community hospitals (52), and commercial laboratories (18). Few (14.6%) were novices with less than 1 year of experience in cytopathology, 19.7% had up to 5 years of experience, and a majority of 65.7% declared an experience of 6 to more than 15 years. The average number of lung cytology specimens seen per week was 11.7 ± 15.5 (0–120).

Quiz Results

The performance of the 3 participant categories (cytopathologists/cytotechnicians/surgical pathologists) was not significantly different and was independent of sex, years of experience, and number of lung cytology specimens examined per week, when all cases were evaluated together. The percentage of correct answers given by cytopathologists, cytotechnicians, and surgical pathologists was 75.3, 70.8, and 67.9% for the control cases but only 32.9, 29.7, and 29.1% for the equivocal cases, respectively. Overall, the control cases were assessed significantly better than were the equivocal cases (P < 0.001). When the 20 participants with the best individual results (14 with more than 5 years of experience; mean, 12.0 specimens per week) and the 20 participants with the lowest number of correct answers (10 with more than 5 years of experience; mean, 9.9 specimens per week) were compared, it became evident that both groups experienced difficulties with the same quiz cases. The 20 best individuals were not significantly better than chance (P = 0.2) in assessing the 5 most difficult quiz cases, i.e. those 5 cases with the highest percentage of inaccurate answers (Cases 6, 7, 15, 19, and 24). Interestingly, the 20 most experienced participants (>15 years of experience, ≥15 specimens per week) performed by no means better at the 5 most difficult cases (Cases 6, 7, 9, 16, and 24) than did the 20 least experienced participants (<1 year of experience). There was no significant difference in the mean number of correctly and approximately correctly solved cases between the most experienced and the least experienced participants (mean, 0.65 ± 0.6 and 1.8 ± 0.8 vs. 0.6 ± 0.7 and 2.35 ± 1.4, respectively; P = not significant). In contrast, the most experienced participants diagnosed the 25 remaining cases significantly more often correctly or approximately correctly than did the least experienced participants (mean, 12.0 ± 3.6 and 17.5 ± 2.8 vs. 10.0 ± 3.3 (P = 0.011) and 15.5 ± 3.1 (P = 0.006), respectively).

Equivocal cytomorphology resulted more often in a false-positive than a false-negative diagnosis. 5 of 8 cytologic specimens of benign cases were misjudged as positive, while only 2 of 15 malignant cytologic specimens were diagnosed as negative by at least 15% of all participants. The 7 control cases were assessed approximately correctly by 87% of the participants, while this percentage fell to 56% for the 23 equivocal cases (P < 0.001). To determine the specific type of benign cells that gave rise to false-positive diagnoses in the category of equivocal cytology, all images of these 8 cases were reviewed by 3 cytopathologists (LB, PD, and PS). The consensus diagnoses were reactive Type II pneumocyte hyperplasia (Cases 2, 7, 9, and 24) and reparative change (Cases 1, 15, 16, and 30). In 2 of these cases, a precise distinction of reparative change and early phase of squamous metaplasia was also discussed (Cases 1 and 16). The two benign controls (Cases 10 and 14) were considered to show normal ciliated cells and some macrophages.

The percentage of undetermined answers was significantly higher in the equivocal cases than in the control cases (P < 0.001, Table 2). The highest quiz-score was achieved by a pathologist practicing at a Swiss university hospital who had between 6 and 10 years of experience and evaluated 18 lung cytologic specimens weekly. He had given 20 correct and 23 approximately correct answers. The average score of all participants was considerably lower (11.7 correct and 18.4 approximately correct answers).

Table 2. Level of Uncertainty and Results Stratified by Diagnostic Group: Mean Percentage of Participants Who Expressed Uncertainty by Giving an Indeterminate Answer for Cases of Different Diagnostic Categories*
No. of CasesDiagnostic froupAnswer categoryResults
  • Abbreviations are explained in the second footnote to Table 1.

  • *

    That is, percentage of participants opting for either of the answer categories favor malignant, favor benign, or unspecified. The column on the far right shows the average percentage of indeterminate, correct, and wrong answers for cases of the different diagnostic groups. Compared with the control cases, inaccurate answers were significantly more frequent for equivocal cases (P < .001). Participants showed a significantly higher level of uncertainty in equivocal cases than in control cases (P < .001).

Favor benign, %Favor malignant, %Unspecified, %Total undetermined, %Correct answer, %Wrong answer, %
2Benign controls11.04.83.719.574.85.7
5Malignant controls4.519.03.226.767.06.3
8Benign equivocal25.918.112.056.020.823.2
13NSCLC equivocal15.423.712.051.134.414.5
2SCLC equivocal9.924.59.944.336.119.6

Results for Selected Cases

Case 6: This FNA specimen of an adenocarcinoma was correctly considered malignant by 39.4% of the 137 participants but misclassified as small-cell lung cancer (SCLC) by 29.9%.

Case 26: Conversely, this SCLC was incorrectly classified as non–small-cell lung cancer (NSCLC) by 17.5% of the participants. The remaining 18 cancers were misclassified by less than 10%.

Case 7: An FNA specimen from a patient with pulmonary hamartoma, a common source of false-positive diagnoses, was considered as NSCLC by 19.0% (Fig. 1).

Figure 1.

Case 7: This FNA specimen with markedly atypical cells (A) from a patient with pulmonary hamartoma was considered as NSCLC by 19.0% of the participants. Biopsy revealed the typical morphology of a pulmonary hamartoma consisting of hyaline cartilage admixed with epithelial and stromal elements (B).

Case16: Only a third of the participants correctly recognized this solitary pulmonary nodule from a patient with tuberculosis as a benign reactive lesion. Another 31.4% misclassified it as NSCLC (Fig. 2).

Figure 2.

Case 16: Reactive bronchial cells with vesicular nuclei and large eosinophilic nucleoli had misled a third of the participants towards a false-positive diagnosis in this case of lung tuberculosis presenting as a solitary pulmonary nodule.

Case 19: Inferior image quality resulted in the highest number of participants expressing uncertainty (70.1%) by opting for either one of the indeterminate answer categories atypical unspecified (×36), favor benign (×32), or favor malignant (×28).

Case 24: This bronchoalveolar lavage specimen of a patient with hemoptysis and a biopsy finding of pulmonary capillaritis resulted in the poorest performance, with only 11.7% participants giving an approximately correct answer and 62.8% diagnosing a NSCLC.

Case 29: This metastatic breast cancer with little atypia was misclassified as benign by 40.9% (Fig. 3).

Figure 3.

Case 29: Metastatic breast cancer cells showing little atypia(A). The nuclei of normal bronchial cells next to the tumor cell cluster are nearly indistinguishable from the breast cancer cells. Immunocytochemically, the metastatic cells are strongly positive for the estrogen receptor (B). Histologically, atypia of intralymphatic tumor cell clusters just beneath the bronchial epithelium is minimal (C) and explains why 40.9% of all participants made a false-negative diagnosis.

DISCUSSION

Cytomorphologic features combined with appropriate clinical information often allow accurate diagnoses of lower respiratory tract cytologic specimens.10, 11 However, equivocal lung cytology is a recurring problem in every cytology laboratory.1–5

In a recently published study of 54 equivocal lung cytology specimens, we demonstrated the diagnostic utility of multitarget FISH in these challenging cases.5 A subset of specimens of this previous study was selected for the present investigation. To determine and make aware of limitations of cytomorphology in diagnosing challenging cases, we constructed an online gallery featuring cell groups that had been analyzed by FISH using a relocalization software.

As expected, participants performed better at control cases, with up to 98.6% making an accurate differentiation of benign from malignant. The relatively low performance for individual control cases may be explained by the fact that only 19% of the participants were fully trained cytopathologists. However, the equivocal pulmonary cytologic specimens overstrained even the 20 individuals with the best quiz performance. They were not able to assess the 5 most difficult cases better than if they had chosen their answers by chance.

Several previous studies have addressed the correlation between remote cytologic diagnoses (telecytology) and diagnoses made by direct examination of the glass slides.12–14 Overall interobserver agreement for the digital diagnoses was fair to excellent. Alli et al.13 found the disagreement between pathologists' glass slide and digital diagnoses of cytologic specimens to be less than interobserver disagreement of pathologists reviewing glass slides. The recent introduction of high quality virtual slides15–17 will further contribute to the reduction of diagnostic inaccuracies of telecytology.

Well-known pitfalls had resulted in poor performance in the equivocal quiz cases, including reactive Type II pneumocyte hyperplasia and bronchial repair. Features that help to distinguish reactive Type II pneumocytes from bronchioloalveolar adenocarcinoma (BAC) have previously been described in detail.2, 4, 18 Nevertheless, it has been conceived that because of overlap of features, reactive Type II alveolar cells cannot always be differentiated from BAC with certainty.2, 4 In our study, reactive atypia of Type II pneumocytes led to a high percentage of false-positive answers in the case of a pulmonary capillaritis (Case 24).

Similarly, some features help to distinguish bronchial repair from NSCLC.4 In bronchial repair the cells form two-dimensional cohesive sheets that contrast with the dyscohesive cells or cell groups in malignancy, as illustrated in Case 16. The nuclei are enlarged but retain regular nuclear membranes. Importantly, the prominent and irregularly shaped nucleoli should not be overinterpreted as evidence of malignancy.19 Granulomatous diseases (Case 16) are a common cause of repair that can give rise to a false-positive diagnosis.3, 4

Another common source of false-positive diagnosis, pulmonary hamartoma, was misclassified as NSCLC by 19% of the participants. This number comes close to the 22% false-positive rate for FNAs of pulmonary hamartomas in a study with 766 participants.20

A multivariate analysis revealed tumor grade (low atypism) as the main cause of diagnostic failure of cytology in breast cancer.21 Accordingly, 40.9% of the participants made a false-negative diagnosis in the case of a metastatic breast cancer with little atypia (Case 29).

Altogether, false-positive quiz diagnoses were more frequent than false-negatives. Although our study highlights sources of false-positive diagnoses, it is important to realize that some of the false-positives by the participants might be related to the setting of this study. It is possible that looking at digital images in a research setting provokes more diagnostic aggressiveness than in the real world of diagnostic routine with regular smears.

It has to be emphasized that the quiz cases comprised a collection of very difficult cases, including some well-known pitfalls. The poor results for selected equivocal quiz cases may not be fully representative of the performance of cytology experts in daily routine diagnostics based on whole smears. However, it illustrates the limitations of cytomorphology in a small subgroup of lesions that may be relevant for the development of computer-assisted screening methods or telecytology applications wherein only single images and not fully digitized slides are available. A cytological diagnosis is a synthesis of all morphological information on a slide. These include the comparison with benign reactive cells and acellular background which are not well represented on single images. Some equivocal cases were a hurdle even for the best of the participants. It is questionable whether educational efforts will enhance the diagnostic success in such challenging cases. However, equivocal cases may nevertheless exert an educational effect by sensitizing (cyto-) pathologists to the limitations of cytomorphology in exceptional cases. A high awareness of possible pitfalls is a prerequisite for avoiding mistakes. Indeed, a higher percentage of indeterminate answers given for equivocal specimens demonstrates the ability of morphologists to identify such cases. This means that (cyto-) pathologists are able to recognize situations in which FISH or other diagnostic adjuncts may be indicated and in which a telecytologic diagnosis on single images should be avoided.

In our experience, FISH coupled with automated relocalization appears to be highly useful to avoid a false-positive and false-negative diagnosis in case of equivocal respiratory cytology. Given the limited sensitivity of this multitarget FISH assay, below 80%, and reports of some false-positive results, FISH should always be discussed in concert with all morphologic and clinical findings.5, 7, 9, 22

Online questionnaires give the participant a good opportunity to benchmark their performance with a large number of international (cyto-) pathologists/cytotechnicians. By introducing participant identification, educational Web sites can additionally provide users with continuing education credit points,23, 24 which doctors and cytotechnologists need to document in many countries. Self-instructional tests using web-based technology may assist in the worldwide harmonization of pathology training and examination, helping to improve the overall quality and status of the specialty.25, 26 Several good examples of similar tests for self-evaluation in cytology23, 27 and pathology are freely accessible in the Internet.24, 28 The positive response of the participants of the lung cytology quiz has inspired us to conduct a similar one on urinary cytology.29

In conclusion, it may be said that the results of the quiz illustrate limitations and common diagnostic problems of lung cytology. The difficulties in equivocal lung cytology that were experienced even by individuals with a high level of experience emphasize the diagnostic utility of molecular adjuncts such as multitarget FISH. Online continuing education is independent of time and location, thereby satisfying the needs of a growing number of users with limited time at their disposal. This is reflected by the overwhelming majority of quiz participants (97%) who would appreciate similar cytopathology courses on the Internet on a regular basis.

Ancillary