Fax: (617) 738-1450
Sentinel lymph node mapping and molecular staging in nonsmall cell lung carcinoma†
Article first published online: 29 AUG 2005
Published 2005 by the American Cancer Society
Volume 104, Issue 7, pages 1453–1461, 1 October 2005
How to Cite
Pulte, D., Li, E., Crawford, B. K., Newman, E., Alexander, A., Mustalish, D. C. and Jacobson, D. R. (2005), Sentinel lymph node mapping and molecular staging in nonsmall cell lung carcinoma. Cancer, 104: 1453–1461. doi: 10.1002/cncr.21325
This article is a U.S. Government work and, as such, is in the public domain in the United States of America.
- Issue published online: 17 SEP 2005
- Article first published online: 29 AUG 2005
- Manuscript Accepted: 25 APR 2005
- Manuscript Revised: 28 MAR 2005
- Manuscript Received: 30 DEC 2004
- National Institutes of Health Research Training. Grant Number: 2T32 CA09454
- American Cancer Society/New York University Institutional Pilot Project Funds
- sentinel lymph node;
- molecular staging;
- nonsmall cell lung carcinoma;
- cytokeratin 7;
Lymph node (LN) involvement predicts recurrence in patients who have undergone resection of apparently localized nonsmall cell lung carcinoma (NSCLC). Standard detection methods for LN disease have a low sensitivity, and many patients with apparent N0 disease status develop recurrent disease. Molecular techniques can improve the detection of micrometastases, whereas sentinel lymph node (SLN) mapping can indicate which LN may contain micrometastases. These methods, although potentially complementary, have not, to the authors' knowledge, been used together previously.
The authors used SLN mapping and molecular staging to improve the detection of LN micrometastases in patients with NSCLC. Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis for cytokeratin-7 (CK7), expressed both in normal lung and in malignant lung, was used to identify tumor-derived material in LN.
SLN mapping was performed in 13 patients, with 1–3 SLNs identified in each patient, and sufficient RNA for RT-PCR was obtained in 12 of these 13 patients. Eleven of 12 tumors expressed CK7. Overall, 32 LNs were positive for CK7, including 13 of 21 SLNs. Ten of 11 patients with evaluable SLNs had at least 1 CK7-positive SLN. Routine pathology showed Stage I disease in eight patients, T3N0 disease in one patient, and LN-positive disease in two patients. Of the nine patients with N0 disease according to routine pathology that was evaluable by RT-PCR, eight patients were upstaged by this technique. All patients with positive LN status by routine pathology who were evaluable by RT-PCR analysis had positive RT-PCR results.
LN micrometastases were common in resected NSCLC, including patients with N0 disease according to routine pathology. SLN mapping was useful for identifying disease-containing LNs. This approach may be useful for stratifying histologically N0 patients into higher risk and lower risk groups. Cancer 2005. Published 2005 by the American Cancer Society.
The mainstay of treatment for patients with early-stage nonsmall cell lung carcinoma (NSCLC) is surgical resection. However, recurrence after surgery is common. Even in patients with Stage IA (T1N0) NSCLC, the recurrence rate after complete resection is 30–40%, and it is well over 50% when lymph node (LN) involvement is detected.1, 2 Undetected metastatic disease is the cause of these recurrences. Better methods for detecting LN involvement would improve the ability to determine the risk of recurrence, which may affect patient treatment, because patients who are at greater risk are more likely to benefit from adjuvant therapy.
LN metastases may remain undetected for two reasons: 1) LN drainage does not always follow the predicted pattern, and 2) methods used to detect tumor-derived material in LNs are insensitive. Previous studies that aimed at better methods for detecting LN micrometastases sought to improve LN identification during surgery through sentinel LN (SLN) mapping or to improve the sensitivity of detecting tumor cells within resected LNs. We are not aware of any previous studies, however, that combined SLN mapping and molecular staging, despite the potential complementarity of these techniques. In the current study, we combined SLN mapping and molecular staging in patients with NSCLC.
Lymph drainage patterns can be established by SLN mapping, in which a lymphophilic dye, a radiolabelled colloid, or both are injected at the edges of a primary tumor, and the dye is tracked to the first LN or the first few LNs encountered, termed the SLN. Aberrant drainage patterns, in which the SLN is not in the predicted anatomic location, are well established for some malignancies, e.g., melanoma and breast carcinoma.3–5
In NSCLC, a few studies using SLN mapping and light microscopy identified ≥ 1 SLN in > 80% of patients and found that the SLN status was predictive of the status of all other LNs in 80–100% of patients.6–10 The main advantage of SLN mapping in NSCLC, if the LNs are examined only by routine histology, is the ability to identify aberrant drainage. Then, these LNs can be resected, improving the accuracy of staging and the chance that all disease has been resected.
Several groups have reported that NSCLC can drain directly to N2 LNs without N1 LN involvement. Such so-called “skip lesions” confer a prognosis more like N1 disease than N2 disease,11–13 and SLN mapping may lead to a more accurate determination of prognosis in such patients. In addition, in patients who have aberrant lymphoid drainage, the true SLN may not be resected during surgery, leading to a pathologic stage of N0 when LN disease actually was left behind at surgery.
SLN mapping not only can improve staging when routine microscopy is used for staging; but, perhaps more important, it can direct the pathologist to those LNs most likely to contain micrometastases. Then, these LNs can be studied by more sensitive means that would be impractical to perform routinely on all resected LNs.
The most often reported method of identifying micrometastases has been staining slides of LNs with immunohistochemistry (IHC) stains followed by light microscopy. In studies of esophageal, gastric, and breast carcinomas,14–17 and in a few NSCLC studies,18–22 LN micrometastases that were overlooked by routine pathology but detected by IHC predicted a worsened prognosis. A few studies, in contrast, failed to demonstrate a prognostic significance to LN micrometastases detected by IHC.23, 24 One way to improve light-microscopic detection of LN micrometastasis is through serial LN sectioning,25, 26 but this labor-intensive approach is impractical outside of the research setting.
Polymerase chain reaction (PCR)-based molecular assays that were developed to detect transcripts of genes expressed by tumor cells, but not by lymphoid tissue, or to detect somatic DNA changes in tumor DNA that are not present in germline DNA, offer a sensitive means of detecting LN micrometastasis.27–29 LN micrometastasis detected by reverse transcriptase-PCR (RT-PCR) has been associated with a worse prognosis in patients with melanoma and prostate carcinoma.30–32 An advantage of molecular staging is that nucleic acid can be isolated from a large fraction of each LN under study, whereas standard light microscopy enables visualization of only a small fraction of each LN. In the current study, we combined the use of SLN mapping with molecular staging by RT-PCR using cytokeratin-7 (CK7), which is a component of normal lung tissue that is expressed by most NSCLCs but not by normal LNs. We also examined LNs by IHC.
MATERIALS AND METHODS
Institutional Review Board approval was obtained for the study, and informed consent was obtained from each patient. Patients for whom surgery was planned for one of two indications were offered participation: 1) biopsy proven NSCLC or 2) lesions suspicious for NSCLC on imaging studies, when surgery was planned for diagnosis and/or treatment.
Each tumor was injected during surgery with a total of 0.5–2.0 cc lymphazurin blue dye in 3–5 areas, usually at 4 quadrants. The first LN or the first few LNs that took up the dye were resected and labeled as the SLN. Each SLN was bisected in the pathology laboratory; half was used for routine examination, and half was snap frozen to − 80 °C and saved for further analysis. The tumor and other LNs were resected in the standard manner. Each non-SLN was processed using the same method that was used to process the SLN. A piece of the primary tumor also was frozen for molecular studies. All frozen specimens were held until the final pathology report was issued in case the tissue was needed for routine analysis. In all samples, the half of each LN that was sent for routine pathology was sufficient, and the frozen half was released for molecular studies.
RNA was isolated using the RNeasy extraction kit (Qiagen, Chatsworth, CA). Tumors and LNs were processed on different days to minimize the risk of cross contamination. No contamination of a negative control was seen.
Using “drop-in, drop-out multiplex” RT-PCR, CK7 and housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were amplified together in 1 reaction.33 This method permits simple (1-tube), yet highly sensitive detection of a gene that is expressed at a low level (in this case CK7) by nested RT-PCR with simultaneous but less sensitive detection of a highly expressed housekeeping gene by nonnested RT-PCR. This method is useful, because the LN RNA is expected to contain few CK7 transcripts on a background of many more GAPDH transcripts.
Briefly, the assay employs 6 primers: 4 primers for CK7 and 2 primers for GAPDH (Fig. 1). The primers used were (5′ to 3′) CK7-1, gctggccccgctgcggctggacgccg; CK7-2, ggacccgcactgctggagaagctcagggc; CK7-3, ggagagcgagcagatcaa; CK7-4, gaactcatcacagagatattc; GAPDH-1, aaggtcggagtcaacggatt; and GAPDH-2, catggactgtggtcatgagt. In Phase I of the reaction, the annealing temperature is high (69 °C) for 20 cycles, permitting binding only of the outer pair of CK7 primers (CK7-1 and CK7-2), which, because of their length (26 base pairs [bp] and 29 bp) and high GC content, have a high maximum temperature (Tm). During Phase II of the reaction (45 cycles), the lower annealing temperature, 56 °C, permits binding both of the nested CK7 primers to the Phase I CK7 RT-PCR product and of the GAPDH primers to the GAPDH transcript. Thus, during Phase I, the low-level CK7 transcripts are amplified without competing with the more abundant GAPDH transcripts for reaction components. Only after the CK7 transcripts have been amplified for 20 cycles is the housekeeping gene amplified, thereby increasing the assay sensitivity for CK7.33
We sought to detect very low-level RNA transcripts, because we predicted that CK7 RNA would be present in only a few copies in the total volume of isolated RNA. Therefore, our results are subject to statistical variation, depending on whether the aliquot taken contained any RNA transcripts. This phenomenon has been described previously for PCR analysis of very dilute samples and must be approached by analyzing each sample multiple times and setting an arbitrary cut-off values for considering a sample positive.34, 35 Therefore, each sample was initially assayed twice; and, if the results agreed, then they were considered confirmed. If the results from two assays were different, then the sample was assayed at least four times. A sample was scored as positive only if it was positive in at least three of four assays. Positive controls (RNA from a known positive cell line or from a previously confirmed tumor) and negative controls (LN RNA from a patient without malignant disease or LN RNA from a patient with colon carcinoma) were included in each experiment. Each assay also contained at least one sample without RNA to control for reagent contamination.
One slide of each tumor and LN was stained for pancytokeratin (antipankeratin primary antibody clones AE1/AE3/PCK26; Ventana Medical Systems, Tucson, AZ),36, 37 and another slide of each tumor and LN was stained for CK7 (CK7 primary antibody clone LP5K; Ventana Medical Systems)38, 39 along with positive and negative controls, except that specimens from Patient 5 were excluded because of an administrative error. RT-PCR and IHC analyses were performed in separate laboratories by different investigators, and investigators were blinded to the results of other studies until all results were obtained.
Samples were obtained from 15 male patients, all current or former smokers, ages 48–79 years (Table 1). One patient underwent pneumonectomy, and the other 14 patients underwent lobectomy. SLN mapping appeared to be successful in 13 patients, although, in 1 patient (No. 8), RNA could not be isolated from the apparent SLN. In two patients (Nos. 3 and 5), SLN mapping was not attempted, once because of macroscopically enlarged LNs that were seen during surgery, and once because the tumor was adherent to the ribs, rendering injection technically difficult.
|Patient||Histologic diagnosis||Successful SLN mapping and RNA isolation||CK7 RT-PCR (positive/total)||Disease stage||DFS (mos)||Current clinical status|
|SLN||Non-SLN||All LNs||SLNs only|
|1||Adeno||Yes||2/3||1/2||T1N0||N0||N2||N1||4||Died of recurrence|
|3||Squamous||No||N/A||1/1||T3N0||N0||N1||N/A||1||Died 36 days after surgery|
|4||Adeno||Yes||2/2||1/1||T2N1||N0b||N2||N2||21||Died of recurrence|
|5||Adenoa||No||N/A||1/2||T3N2||N/A||N2||N/A||11||Died of recurrence|
|8||Large cell||Noc||N/A||2/3||T4N2||N2||N2||N/A||5||Died of recurrence|
|10||Large cell||Yes||1/1||4/6||T2N2||N2||N2||N1||10||Died of recurrence|
|14||Squamous||Yes||1/1||0/0||T3N0||N0||N1||N1||10d||Alive, second primary|
|15||Small cell||Yes||0/1e||N/Ae||Limited||N/A||N/A||N/A||33||Alive, NED|
The only reaction to dye injection was seen in 1 patient (No. 4) who had a transient decrease in O2 saturation, with spontaneous recovery in 2 minutes. Fourteen of 15 patients recovered uneventfully from surgery and were discharged expeditiously from the hospital. One patient on whom SLN mapping could not be performed patient (No. 3) had a long, complicated postoperative course and died on postoperative Day 43 without known tumor recurrence.
One to 3 SLN were identified in each of the 13 patients injected (12 of 13 patients during surgery, and 1 patient [Patient 15] only during pathologic examination) (Table 1). In 9 of 12 patients who had an SLN identified during surgery, identification occurred within 5 minutes of dye injection. In the other 3 patients, identification took 8 minutes, 21 minutes, and 25 minutes, respectively. In the patient for whom SLN identification took 21 minutes, the search for the SLN was delayed due to the oxygen desaturation mentioned above and consequent intraoperative measures. In the other two patients and in the patient who had an SLN identified at pathology, the SLNs were found within the lung (between two lobes) or behind the tumor, hindering their identification.
Routine Histologic Examination
Pathologic examination of the tumors revealed eight adenocarcinomas (including three bronchoalveolar carcinomas), four squamous cell carcinomas, two large cell carcinomas, and one small cell lung carcinoma (SCLC). The latter diagnosis was not known preoperatively.
Routine histology showed LN negative (N0) disease in 10 of 15 patients (Table 1). Eight of those 10 patients Stage I disease (T1N0 or T2N0), and the other 2 patients had T3N0 disease. LNs were positive in five patients, with N1 disease in one patient (adenocarcinoma), N2 disease in three patients (two large cell carcinomas and one bronchoalveolar carcinoma), and limited stage SCLC in one patient.
Two lung carcinoma cell lines (A549 and H358) were positive for CK7 expression, and one lung carcinoma cell line (H460) was negative. Three of four samples from NSCLC biopsies were positive. The glioblastoma cell line U87 and the colon carcinoma cell line SW480 were negative for CK7 expression. Nine negative control LNs from patients without carcinoma or with colon carcinoma, including several that were positive for CK20 (expressed by gastrointestinal epithelium) according to a similar assay, all were negative for CK7.
To test assay sensitivity, RNA was prepared from a known number of cells from the positive lung carcinoma cell line H358, and serial dilutions down to RNA from 10− 1 cells were assayed. The RT-PCR reaction showed a strong band for 10 cells per tube and a faint band for 1 cell per tube (Fig. 2). This sensitivity is similar to that reported previously for this type of assay.33
Fourteen of 15 primary tumors that were examined by RT-PCR (all tumors except the 1 SCLC) were positive for CK7 expression. All four squamous cell carcinomas produced RT-PCR results that were positive for CK7 expression, even though two were negative for CK7 according to IHC analysis, and other investigators also found that squamous cell carcinomas generally are negative for CK7 according to IHC40, 41 (Table 1). These results confirm that RT-PCR for CK7 expression is more sensitive than IHC.
Overall, of the 62 LNs tested by RT-PCR, 59 were tested by both IHC and RT-PCR (Figs. 3, 4). Among the 14 patients who had CK7-expressing tumors, 1 or more LN was positive by RT-PCR in 13 patients. Of the 11 patients who had an evaluable SLN, at least 1 SLN was positive by RT-PCR in 10 patients. In the one man (No. 6) who had all SLNs that were negative by RT-PCR, the two non-SLNs also were RT-PCR-negative. Thus, in all patients, the SLN was predictive of overall LN status.
Patient Upstaging by RT-PCR
Of the eight patients who had Stage I disease according to routine pathology, seven patients were upstaged by RT-PCR (Fig. 5, Table 1). Three patients were upstaged from Stage I to Stage II (from N0 to N1), and four patients were upstaged to Stage IIIA (from N0 to N2). One patient remained in Stage IA after RT-PCR staging. Three patients were upstaged from Stage II to Stage IIIA (one patient moved from T2N1 to T2N2, and two patients moved from T3N0 to T3N1). Three patients had Stage IIIA disease based on both routine pathology and RT-PCR. In this analysis, the RT-PCR results from all LNs studied were considered.
If only the RT-PCR results from the SLN are considered, then the same seven patients with Stage I disease still were upstaged, but they were upstaged to N1 (Table 1). One patient (No. 4) was upstaged from N1 to N2 based on RT-PCR results of the SLN alone. However, the majority of our SLNs (11 of 12) were N1. In the 14 patients who had CK7 RT-PCR-positive primary tumors, all histologically positive LNs also were positive for CK7 by RT-PCR.
Ten of 14 tumors and only 1 of 59 LNs examined stained positive for CK7. All 14 tumors stained positive for pankeratin (AE1/AE3); however, even in these tumors, only 7 of 59 LNs from 5 patients were positive for AE1/AE3, leading to the upstaging of 2 patients (Table 1). All IHC-positive LNs also were positive by RT-PCR.
Although this was intended as a proof-of-principle study and was not powered to distinguish differences in recurrence and survival, follow-up clinical information on the patients was obtained. During a follow-up of 33–44 months, of the 14 patients with NSCLC, 7 patients died, including 5 who died recurrent disease, 1 who died (No. 3) in the postoperative period, and 1 who died from acquired immunodeficiency syndrome (No. 12) without evidence of recurrence. Of the patients who remained alive, 1 patient (No. 14) was treated for an apparent second primary lung carcinoma 10 months after resection. Of the eight patients with N2 disease according to the RT-PCR analysis, five patients developed recurrent disease, including one patient with N0 disease as determined in routine pathology. The primary tumor has not recurred in any patient with N0 or N1 disease by RT-PCR staging.
The results of the current study confirm the feasibility of SLN mapping in NSCLC and demonstrate that molecular studies on the SLN can improve greatly the detection of LN micrometastases in NSCLC. The SLN was identified in < 5 minutes in 9 of 13 patients. The only side effect related to the procedure was a transient, intraoperative decrease in oxygen saturation in one patient. Rapid identification of the SLN in patients with NSCLC should improve as thoracic surgeons become more familiar with the technique, similar to what has occurred in other tumor types.42, 43
The current study data indicate that RT-PCR for CK7 is a highly sensitive marker for NSCLC. The use of a single tube for the entire nested RT-PCR assay decreases the risk of false-positive results that arise from contamination.33 The risk of a false-negative result is minimized by multiple testing of each sample. RT-PCR is not labor intensive so its use should be feasible in the clinical setting, especially if RT-PCR testing is limited to a subset of LNs (the SLN).
Previous studies of micrometastatic disease in patients with NSCLC using RT-PCR on all resected LNs (without SLN mapping) also have found high rates of LN positivity. In 1 study, in which MUC1 was used as a marker, LN micrometastases were found in 6 of 10 patients.44 In another study, RT-PCR for carcinoembryonic antigen was positive in 25% of LNs from patients who had histologic Stage I NSCLC. No mortality or recurrence data were provided by those investigators.45
In the current study, we found RT-PCR evidence of micrometastasis in the SLN in 10 of 11 patients (91%). This high percentage likely resulted from two factors: first, the relative ubiquity of CK7 expression in NSCLC compared with the markers used in other studies and, second, our highly sensitive assay, which uses nested RT-PCR.
In each evaluable patient who had an SLN identified and who had any LNs with positive RT-PCR results, at least one SLN was among the positive LNs; thus, in this study, the SLN was suitable as a surrogate marker for the entire LN basin in defining LN-negative disease versus LN-positive disease. When the SLN is an N1 LN, then the other LNs still must be examined by some means to distinguish between N1 and N2 or N3 involvement. Conversely, with recent developments in NSCLC treatment, distinguishing between N1 and N2 or N3 disease is less important than distinguishing between N0 and N1 disease, because adjuvant chemotherapy is becoming standard for most patients with LN-positive disease. Thus, a practical way that our approach could be applied clinically would be to examine the SLN by using molecular techniques, with other resected LNs examined only by standard histology; alternatively, additional LNs could be examined by using molecular techniques in only the few patients (e.g., Patient 6 in the current series) with all SLNs negative according to the RT-PCR results.
Our current results suggest that tumor-derived material spreads to LNs early in the course of NSCLC. This finding is consistent with results from other studies of micrometastatic disease in LNs by RT-PCR and with the high recurrence rates seen for patients with NSCLC after surgery, even for those patients with histologically N0 disease.
Compared with RT-PCR, IHC enabled the detection of disease that had not been seen by routine hematoxylin and eosin staining in only three LNs from two patients. The much greater sensitivity of RT-PCR may result from the limits imposed on the histologic assays by sampling: RNA was isolated from one-half of each LN, whereas routine and IHC staining was performed on only one or two slides per LN, which contained much less material. Serial sectioning improves the sensitivity of detecting LN micrometastases compared with the examination of fewer sections25 but yields hundreds or thousands of slides per LN, the routine examination of which is not feasible. In addition, in our series, four of the four squamous cell carcinomas had RT-PCR results that were positive for CK7, whereas only two of the four had CK7-positive by IHC results, confirming that the RT-PCR assay is more sensitive than IHC in detecting low-level CK7 expression.
The prognostic significance of molecular upstaging in NSCLC is not yet known. However, it is noteworthy that none of the patients with N0 or N1 disease according to the RT-PCR analysis developed recurrent disease by the time of last follow-up, in contrast to recurrences reported in five of eight patients with N2 disease according to the RT-PCR analysis.
Adjuvant chemotherapy is becoming the standard of care for most patients with LN-positive disease.46, 47 However, in 1 study, 2.9% of patients who received chemotherapy had a Grade 3 toxicity (according to the Japanese Society of Clinical Oncology)46; whereas, in another study, the death rate from chemotherapy was 0.8%, and there was little or no benefit for patients with Stage N0 disease.47 In contrast, another recent study found a benefit to chemotherapy in patients with Stage IB (N0) disease.48 Thus, although accumulating evidence indicates that adjuvant chemotherapy benefits some patients after surgery for NSCLC, its risks are considerable, and the benefit in patients with early-stage disease remains unclear.
Our methodology may be helpful in selecting the best risk patients who are most likely to be cured by surgery alone and therefore are least likely to benefit from adjuvant therapy. This may be a subset of patients who should be spared the toxicity of adjuvant chemotherapy, even as this approach becomes standard for many patients after they undergo resection for NSCLC.
- 45Molecular staging of lung cancer: real-time polymerase chain reaction estimation of lymph node micrometastatic tumor cell burden in Stage I non-small cell lung cancer—preliminary results of Cancer and Leukemia Group B Trial 9761. J Thorac Cardiovasc Surg. 2002; 123: 484–491., , , et al.
- 48Randomized clinical trial of adjuvant chemotherapy with paclitaxel and carboplatin following resection in stage IB non-small cell lung cancer (NSCLC): report of Cancer and Leukemia Group B (CALGB) Protocol 9633 [abstract]. J Clin Oncol. 2004: 22(14S): A7019., , , et al.