Correspondence

Detection of Carcinoembryonic Antigen Messenger RNA in Lymph Nodes from Patients with Colorectal Cancer

N Engl J Med 1998; 339:1642-1644November 26, 1998

Article

To the Editor:

The study by Liefers et al. (July 23 issue)1 demonstrated the prognostic importance of micrometastases detected by the use of a tissue-specific reverse-transcriptase–polymerase-chain-reaction assay (PCR) in patients with colorectal cancer. However, certain issues are of concern. Only 192 lymph nodes were analyzed, an average of 7 per patient. Histologic assessment of lymph nodes in stage T3 colorectal cancer by Goldstein et al. has shown the importance of the number of nodes examined per specimen.2 When 5 to 8 nodes were examined, only 31 percent of the specimens were found to have metastases, but when 17 to 20 nodes were examined, 87 percent of the specimens were found to have metastases. These findings suggest that the true incidence of micrometastases may have been underestimated, but we have found that the likelihood of detecting micrometastases in histologically negative nodes increases when metastases are present in other lymph nodes.3

Sample size is important in determining sensitivity and specificity. Only seven lymph nodes from two patients were used by Liefers et al. as negative controls. Although there was no detectable carcinoembryonic antigen messenger RNA (mRNA) in the negative controls (a specificity of 100 percent), the sample size was such that the 95 percent confidence interval for the specificity of the assay was 15.8 to 100 percent. The sensitivity of this assay can be determined only by including nodes from patients with stage III colorectal cancer, since a comparison with histologically detected lymph-node metastases remains the gold standard.

Arend E.H. Merrie, M.B., Ch.B.
Kankatsu Yun, M.D., Ph.D.
University of Otago, Dunedin, New Zealand

John L. McCall, M.D.
University of Auckland, Auckland, New Zealand

3 References

1. 1

   Liefers G-J, Cleton-Jansen A-M, van de Velde CJH, et al. Micrometastases and survival in stage II colorectal cancer. N Engl J Med 1998;339:223-228
   Full Text | Web of Science | Medline

2. 2

   Goldstein NS, Sanford W, Coffey M, Layfield LJ. Lymph node recovery from colorectal resection specimens removed for adenocarcinoma: trends over time and a recommendation for a minimum number of lymph nodes to be recovered. Am J Clin Pathol 1996;106:209-216
   Web of Science | Medline

3. 3

   Merrie AEH, Gunn J, Phillips LV, Yun K, McCall JL. The anatomical distribution of colorectal cancer lymph node micrometastases. Aust N Z J Surg 1998;68:Suppl:A135-A135 abstract.
   

To the Editor:

Liefers et al. reported a significant correlation between a positive reverse-transcriptase–PCR result and decreased survival, but they did not assess the sensitivity and specificity of their procedure. To determine sensitivity and reproducibility, carcinoembryonic antigen–producing carcinoma cells (from the COLO 357 cell line) should be mixed with mononuclear cells to produce serial 10-fold dilutions.1 As described by Gerhard et al.,1 these mixtures should be subjected to 10 independent reverse-transcriptase–PCR amplifications for carcinoembryonic antigen mRNA. One cannot assume that the limit of sensitivity for the test used by Liefers et al. is identical to that reported by Gerhard et al., because Liefers et al. used different PCR conditions (15 instead of 30 cycles in the second round of reverse-transcriptase PCR).

Another issue is the small number of control lymph nodes tested (7 lymph nodes from only two patients) — clearly a smaller number than was used by Mori et al. (25 lymph nodes) to assess the specificity of reverse-transcriptase PCR for carcinoembryonic antigen mRNA.2 The need for a large number of control samples is also indicated by the false positive results when Liefers et al. used 20 or 25 cycles in the second round of PCR amplification. These false positive results could be due to the process of illegitimate transcription (i.e., transcription of any gene in any cell type).3 Although the estimated number of these transcripts in inappropriate cells is very low (1 mRNA molecule per 100 to 1000 cells), it can cause false positive results because of the high sensitivity of reverse-transcriptase PCR.4 I believe that the use of reverse-transcriptase PCR to detect micrometastases has the potential to improve the stratification of patients with solid tumors.4 However, the sensitivity and specificity of such tests should be accurately assessed before they are used in the clinical setting.

Ronald A. Ghossein, M.D.
Memorial Sloan-Kettering Cancer Center, New York, NY 10021

4 References

1. 1

   Gerhard M, Juhl H, Kalthoff H, Schreiber HW, Wagener C, Neumaier M. Specific detection of carcinoembryonic antigen-expressing tumor cells in bone marrow aspirates by polymerase chain reaction. J Clin Oncol 1994;12:725-729
   Web of Science | Medline

2. 2

   Mori M, Mimori K, Ueo H, et al. Clinical significance of molecular detection of carcinoma cells in lymph nodes and peripheral blood by reverse transcription-polymerase chain reaction in patients with gastrointestinal or breast carcinomas. J Clin Oncol 1998;16:128-132
   Web of Science | Medline

3. 3

   Kaplan J-C, Kahn A, Chelly J. Illegitimate transcription: its use in the study of inherited disease. Hum Mutat 1992;1:357-360
   CrossRef | Medline

4. 4

   Ghossein RA, Rosai J. Polymerase chain reaction in the detection of micrometastases and circulating tumor cells. Cancer 1996;78:10-16
   CrossRef | Web of Science | Medline

To the Editor:

The findings reported by Liefers et al. confirm reports of the prognostic importance of occult regional metastases in patients with colon cancer,1 but there are several problems with their approach. They assumed that carcinoembryonic antigen mRNA is expressed in colon cancer but not in normal lymphocytes, claiming that carcinoembryonic antigen is “cancer-specific” and not expressed in normal tissues. This notion contradicts data from numerous studies showing that carcinoembryonic antigen is expressed in normal epithelial and nonepithelial tissues.2,3 To detect carcinoembryonic antigen reverse-transcriptase–PCR products, Liefers et al. used ethidium bromide gel electrophoresis, which is a subjective and controversial method, particularly when there are faint bands. The authors even indicated that carcinoembryonic antigen mRNA in normal lymph nodes was detected by increasing PCR amplification by 5 or 10 cycles. Reverse-transcriptase–PCR products should be verified by hybridization with a specific probe.

Recent studies have demonstrated that carcinoembryonic antigen mRNA is frequently present in normal lymph nodes, bone marrow cells, and blood cells from patients without cancer.2,3 The limitations of reverse-transcriptase PCR with the primers described by the authors and ethidium bromide gel electrophoresis as compared with Southern blot analysis are shown in Figure 1AFigure 1Reverse-Transcriptase–PCR Complementary DNA Assessed by Ethidium Bromide Gel Electrophoresis as Compared with Southern Blot Analysis with a Carcinoembryonic Antigen–Specific Probe. and Figure 1B. Figure 1C shows examples of RNA from normal lymph nodes and blood cells analyzed by reverse-transcriptase PCR; the results show that the presence of carcinoembryonic antigen mRNA is nonspecific.

Carcinoembryonic antigen mRNA in lymph nodes may come from normal epithelial cells sloughed off into draining lymph nodes or by low-level production of carcinoembryonic antigen mRNA in normal lymph-node cells. Although occult regional lymph-node metastases can be detected in patients with colon cancer,1 it is unwarranted to assume that a positive PCR signal for carcinoembryonic antigen indicates the presence of occult metastasis without further evidence. Nevertheless, it is noteworthy that the authors showed, with the use of ethidium bromide, that patients with positive lymph nodes on reverse-transcriptase–PCR testing have an increased recurrence rate as compared with patients with no detectable carcinoembryonic antigen mRNA.

Peter J. Bostick, M.D.
David S.B. Hoon, Ph.D.
John Wayne Cancer Institute, Santa Monica, CA 90404

Richard J. Cote, M.D.
University of Southern California School of Medicine, Los Angeles, CA 90033

3 References

1. 1

   Greenson JK, Isenhart CE, Rice R, Mojzisik C, Houchens D, Martin EW Jr. Identification of occult micrometastases in pericolic lymph nodes of Dukes' B colorectal cancer patients using monoclonal antibodies against cytokeratin and CC49: correlation with long-term survival. Cancer 1994;73:563-569
   CrossRef | Web of Science | Medline

2. 2

   Bostick PJ, Chatterjee S, Chi DD, et al. Limitations of specific reverse-transcriptase polymerase chain reaction markers in the detection of metastases in the lymph nodes and blood of breast cancer patients. J Clin Oncol 1998;16:2632-2640
   Web of Science | Medline

3. 3

   Zippelius P, Kufer P, Honold G, et al. Limitations of reverse-transcriptase polymerase chain reaction analyses for detection of micrometastatic epithelial cancer cells in bone marrow. J Clin Oncol 1997;15:2701-2708
   Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: The reverse-transcriptase–PCR assay for carcinoembryonic antigen has been well documented and described.1,2 Our report relates the results with this assay to the survival of patients with colorectal cancer. Detection of submicroscopical amounts of tumor cells by reverse-transcriptase PCR for carcinoembryonic antigen is therefore not only of academic value but also of practical value.3

The points noted by Merrie et al., Ghossein, and Bostick et al. are valid. Our series was too small to determine specificity and sensitivity accurately with small confidence intervals. The false positive results when the number of PCR cycles was increased were due to illegitimate expression or carcinoembryonic antigen expression by noncancerous gut epithelial cells in lymph nodes. This phenomenon, also reported by Mori et al.,2 limits the sensitivity of the assay. The consistently negative results with our control nodes (coamplified in every round of PCR) indicate that a signal found under our conditions may be considered “cancer-specific.” The actual number of control nodes we analyzed was arbitrary. Dilution experiments with the use of cultured cells do not realistically reflect the sensitivity of the assay, since the PCR conditions and the quality of the RNA may differ substantially among clinical samples.

Visualization of PCR products both by ethidium bromide gel electrophoresis and by hybridization with a specific probe can be subjective, especially with faint bands. The choice of a method to use in a clinical setting will be determined by its sensitivity, but even more by its potential to discriminate between true micrometastases and background noise. In our study, sequence analysis showed that part of the carcinoembryonic antigen gene was indeed amplified, and all products were analyzed against the background of negative controls.

If our results are confirmed by others, the logical next step will be to determine whether patients with high-risk stage II disease will benefit from adjuvant therapy. This will require a large, randomized, multicenter trial in which the assay used to detect micrometastases gives unequivocal, reproducible results and is easy to use at all participating centers. Our study indicates that it is worthwhile to design such a “consensus test” and use it in a clinical trial.

Gerrit-Jan Liefers, M.D.
Anne-Marie Cleton-Jansen, Ph.D.
Rob A.E.M. Tollenaar, M.D., Ph.D.
Leiden University Medical Center, 2300 RC Leiden, the Netherlands

3 References

1. 1

   Gerhard M, Juhl H, Kalthoff H, Schreiber HW, Wagener C, Neumaier M. Specific detection of carcinoembryonic antigen-expressing tumor cells in bone marrow aspirates by polymerase chain reaction. J Clin Oncol 1994;12:725-729
   Web of Science | Medline

2. 2

   Mori M, Mimori K, Inoue H, et al. Detection of cancer micrometastases in lymph nodes by reverse transcriptase-polymerase chain reaction. Cancer Res 1995;55:3417-3420
   Web of Science | Medline

3. 3

   Liefers G-J, Cleton-Jansen A-M, van de Velde CJH, et al. Micrometastases and survival in stage II colorectal cancer. N Engl J Med 1998;339:223-228
   Full Text | Web of Science | Medline

Citing Articles (2)

Citing Articles

1. 1

   Peter Kienle, Moritz Koch. (2001) Minimal residual disease in gastrointestinal cancer. Seminars in Surgical Oncology 20:4, 282-293
   CrossRef

2. 2

   Kankatsu Yun, Arend E. H. Merrie, Jeremy Gunn, Lynn V. Phillips, John L. McCall. (2000) Keratin 20 is a specific marker of submicroscopic lymph node metastases in colorectal cancer: validation by K-RAS mutations. The Journal of Pathology 191:1, 21-26
   CrossRef