A prospective study including consecutive patients with diagnosis or high suspicion of non-small cell lung carcinoma, in which EBUS-guided needle cytology of mediastinal or hilar lymphadenopathy was performed for diagnosis and/or staging purposes. Case inclusion was determined by the multidisciplinary LC committee of the center (www.chuvi.es), according to the local protocol for diagnosis, staging and treatment of lung cancer which follows the current national and international standards and recommendations. Patients with hilar and/or mediastinal lymphadenopathy with a short axis diameter of greater than 1cm on computed tomography, or 5mm in case of pathological uptake in positron emission tomography-computed tomography, were included. Patients without an evaluable sample according to the cytologist, or an insufficient sample for further methylation studies were excluded. All procedures were performed in Bronchopleural Technical Unit of CHUVI Pneumology Service.

The study was conducted according to the clinical and ethical principles of the Spanish Government and the Declaration of Helsinki, and was approved by the Ethics Committee for Clinical Research of Galicia. Informed consent was obtained from all patients and anonymity was guaranteed.

All procedures were performed consecutively between September 2009 and December 2012. Most were ambulatory and took place in a conventional bronchoscopy room with echocardiography, blood pressure and pulse oximetry monitoring, and under conscious sedation with midazolam and fentanyl. A bronchoscope model BF-UC180F-OL8 (Olympus, Tokyo, Japan) and ultrasound equipment Aloka ProSound Alpha 5 (Hitachi-Aloka, Tokyo, Japan) were used. After endoscopic examination, the lesion was measured and punctured using a NA2015X-4022 needle (Olympus, Tokyo, Japan), advancing through the ultrasound bronchoscope canal to the airway, where the needle was extracted and inserted into the node through the tracheal and bronchial wall with ultrasound guidance. In the node the needle was mobilized and vacuum suction with syringe was performed. Depending on the immediate results obtained, one to three passes were made. The material was recovered and the sample was fixed in alcohol and immediately examined by a cytologist, who identified it as follows: normal node (predominantly lymphoid cells without atypia and/or anthracotic material); node with neoplastic infiltration (presence of neoplastic lymph node cells and cellularity); or non-evaluable sample (presence of only blood or bronchial cells). A part of the samples verified in situ as representative was resuspended in sterile saline solution and immediately frozen at −20°C for subsequent molecular studies.

In cases in which the pathology report found no neoplastic involvement, negative results were verified by thoracic surgery. Patients who did not undergo confirmation by surgery were followed through their medical records for a year. If the lymphadenopathy did not grow during the year, the result was considered a true negative.8 All doubtful cases were excluded from the study.

DNA was extracted using the QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA), adapting the protocol to the volume of each sample. DNA was eluted with 50μL of water and two aliquots were separated for further treatment with sodium bisulfite. DNA concentration was measured with NanoDrop 2000c (Thermo Scientific). The 260/280nm absorbance ratios allowed estimating their purity as optimal (ratio 1.7–2.0), good (ratio > 2.0), or presence of proteins or other contaminants (ratio < 1.7). DNA was stored at −20°C until use.

Methylation analysis required treatment of DNA with sodium bisulfite. EZ DNA Methylation Direct Kit (Zymo Research, Irvine, CA, USA) was used for this purpose, following the manufacturer's recommended protocol, and eluting with 20μL. Modified DNA becomes a uracil-rich single chain, chemically similar to RNA, so its concentration was determined using the NanoDrop (Thermo Scientific), as if it were this nucleic acid. The 260/280nm ratios allowed purity to be assessed as optimal (ratio 1.8–2.2), good (ratio >2.2), or with presence of proteins or other contaminants (ratio <1.8). Treated DNA was stored at −80°C until use.

Methylation analysis of p16/INK4a and SHOX2 genes was performed using nested methylation specific (MS) real-time PCR (qPCR), with 3μL of bisulfite-treated DNA. External primers were used in the first round to amplify the CpG island regardless of methylation status. MS-qPCR was subsequently performed by diluting the PCR product and using methylation-specific probes and primers in a StepOne equipment (Applied Biosystems). More detailed information of the sequences used are available on request.

Pyrosequencing was performed for p16/INK4a and SHOX2 genes. PCR was performed with 3μL of modified DNA using primers designed with PyroMark Assay Design v2.0 (Qiagen) software. Sequencing and methylation quantification were then carried out using the pyrosequencing primer in a PyroMark Q24 (Qiagen). The CpG islands studied were the same as those analyzed by qPCR-MS. More detailed information of the sequences used are available on request.

The SPSS package (v19.0) was used for statistical analysis. Two-tailed tests were used, and P-values <.05 were considered significant. DNA concentration was expressed as median and interquartile range (IQR). Normal distribution and homogeneity of variance were tested with Kolmogorov–Smirnov and Levene tests, respectively. Non-parametric analysis for two independent samples was performed using Mann–Whitney U test, while Kruskal–Wallis was used for more than two independent samples. A Wilcoxon test was used to analyze differences in paired data. The variables studied were gender, tumor type, location of lymph nodes, diagnosis of lymphadenopathy, appearance of the nodal sample and concentration and purity of unmodified and sodium bisulfite-modified DNA.

Sample size was estimated using the Granmo 5.2 application (Municipal Institute of Medical Research, 1999) for paired samples, with an alpha risk of 0.05 and a beta risk of 0.20. The estimated prevalence of malignant lymph nodes was assumed to be 70%, and the assumed difference between sensitivity and negative predictive value of molecular techniques and cytohistological determinations was 15% and 10%, respectively. A minimum of 79 cases was estimated as necessary for performing the study.

Eighty-eight patients were included, yielding 170 samples from different lymph node stations obtained by EBUS-TBNA. The patients were distributed as follows: 38 patients with one sample, 29 with two samples, 13 with three samples, 5 with four samples, and 3 with five samples. Seventy-four patients (84%) were male, with a median age of 62.2 years (IQR, 51–73 years). On a proprietary color scale, 44.7% (n=76) of the cytological node samples were of turbid-clear appearance, 38.2% (n=65) were slightly bloody, and 17.1% (n=29) were very bloody.

In 86.5% of cases (n=147) punctures were performed in mediastinal lymph nodes, while 13.5% (n=23) were performed in hilar lymph nodes. The lymphadenopathy most frequently punctured was 4R (39.4%, n=67) and 7 (33.5%, n=57), followed by 4L (9.4%, n=16), 11L (7.1%, n=12), 11R (4.7%, n=8), 8 (2.4%, n=4), 2R (1.8%, n=3), 10R (1.2%, n=2) and 12R (0.6%, n=1). According to the definitive lymph node diagnosis, 48.2% (n=82) were benign (non-metastatic), while the remaining 51.8% (n=88) were malignant (metastatic). Among these, 61.4% (n=54) were adenocarcinoma, 31.8% (n=28) squamous cell carcinoma and 6.8% (n=6) large cell carcinoma.

A wide range of DNA concentrations was obtained, from 4.10 to 1103.00ng/μL (median: 42.75ng/μL; IQR: 22.48–100.13ng/μL). In order to study DNA characteristics, concentrations were related to epidemiological and clinical parameters (Table 1). Males presented higher median values, with significant differences compared to women (P=.037). Differences in DNA concentration were detected depending on the appearance of cytological samples, with higher values for bloody samples, followed by slightly bloody, and finally, turbid-clear appearance (P<.001), according to the cell contents.

The DNA extraction protocol used provided optimal or good purity (ratio 260/280nm) in 61.18% and 8.24% of cases, respectively. Higher DNA concentrations were obtained from optimal purity samples, followed by those with contaminants, and finally, good purity samples (P<.001, Table 1).

Regarding the node station punctured, no differences in DNA concentration were found when comparing hilar and mediastinal lymph nodes (P=.483), nor when lymph nodes mostly from stations 4R and 7 were compared with the rest (P=.851). No differences in the concentration of DNA were observed between lymphadenopathies according to site. However, lymphadenopathies from station 8 had a higher DNA concentration compared to stations 4R, 11R and 11L (Table 1). DNA concentration did not have normal distribution or homogeneity of variance over the diagnostic groups, with medians of approximately 40ng/μL for the benign, adenocarcinoma and squamous cell categories, and 138.4ng/μL for the six cases of large cell tumor (Table 1). Consequently, no differences were detected in DNA concentration in terms of nodal diagnosis (P=.163) or between benign and malignant lymph nodes (P=.390).

The concentration of modified DNA in relation to some variables was similarly studied (Table 1). The concentrations recorded were lower than those observed before treatment, ranging from 0.20 to 392.30ng/μL (median: 19.13ng/μL, IQR: 7.60–60.18ng/μL). Comparing the concentrations before and after treatment, significant differences (P<.001) were observed, with lower concentrations in 93.53% of the samples.

Although men continued to show higher DNA concentrations than women, the differences were not significant (P=.196). Concentrations of modified DNA differed significantly, depending on the appearance of the samples (P<.001). Moreover, the purity estimation of the modified DNA indicated that 63.53% and 26.47% were of optimal and good quality, respectively, with up to 90% of the samples potentially suitable for methylation analysis. The median for optimal purity samples was the highest, at least three times higher than that of the samples of lower purity (P<.001).

Regarding the lymph node station punctured, no difference was detected between the mediastinal and hilar lymphadenopathies (P=.512), nor between stations 7 and 4R compared to the rest (P=.833), as shown in Table 1. Similarly to DNA concentration before treatment, the differences in modified DNA were found in lymphadenopathies 2R, 4R, 11R and 11L compared to those from station 8 (Table 1). Finally, no differences were found when comparing the nodal diagnostic categories (P=.164) or benign and malignant lymph nodes (P=.093) (Table 1).

Although most samples seemed adequate for further studies, validity was confirmed with methylation analysis of two tumor suppressor genes, using two different methodologies. Table 2 summarizes sample characteristics according to the methylation data collected (samples with/without methylation data) for both genes and the two approaches.

The MS-qPCR technique did not allow analysis of 9.41% (16/170; p16/INK4a) and 1.76% (3/170; SHOX2) samples, whereas pyrosequencing provided no information in 1.26% (2/159; p16/INK4a) and 2.63% (1/38; SHOX2) of cases. DNA concentration medians and IQRs before and after treatment were significantly lower in the samples without methylation data compared to those in which these data were available. This difference was statistically significant (Table 2) and related to the difficulty in testing certain samples. However, the percentage of samples with inadequate purity varied widely both for samples without methylation data and for those with data; this was greater in the first group in all cases and was also associated with the impossibility of obtaining results with the techniques used. Fig. 1 shows the qPCR amplification curves for p16/INK4a in a valid and an invalid sample. The pyrograms of p16/INK4a for a valid and an invalid sample are shown in Fig. 2.

Fig. 1.

p16/INK4a amplification curve by qPCR-MS in a valid sample (green line) and an invalid sample (yellow line). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(0.19MB).

Fig. 2.

p16/INK4a pyrograms of a valid sample (top panels) and an invalid sample (bottom panels). Pyrosequencing was performed in two different reactions to cover the 18 CpG sites of the island: pyrograms on the left show the first seven CpG sites, and on the right, the last eleven CpG sites. The gray bars in the pyrograms indicate each CpG site.

(0.49MB).