| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Correspondence |
University of Manitoba, Winnipeg, Manitoba, Canada
INSERM U618, IFR135, Tours Cedex, France
To the Editor-in-Chief:
We have read with great interest the article by Koga et al citing mammaglobin A as a putative marker for differential diagnosis of lung tumors in patients with a history of breast cancer.1 Mammaglobin A (MGB1) was one of the three breast-specific genes identified by these authors following a serial analysis of gene expression (SAGE) database search (http://cgap.nci.nih.gov/SAGE). The others two genes were the small breast epithelial mucin (SBEM) and the prostate epithelium-specific Ets transcription factor (PDEF). Because the lung is the primary site for breast cancer metastasis, Koga et al tested the efficacy of these candidate genes as potential biomarkers in distinguishing primary lung cancers from metastatic breast cancers.1 They reported that neither SBEM or PDEF were useful for differential diagnosis of lung tumors as these two genes are expressed in normal bronchus (21 of 22 samples positive for SBEM, 20 of 22 positive for PDEF). In contrast, they found all normal lung samples negative for MGB1, thereby making it an ideal candidate for identifying breast cancer metastasis to the lung.1
It is of particular interest to note that the results of their analysis of the SAGE libraries showed that the ratio of total numbers of breast Tags to lung Tags in normal tissues was higher for MGB1 (9:2) compared to SBEM (376:0).1 Furthermore, Tags corresponding to MGB1 have been positively identified in a normal lung library whereas Tags corresponding to SBEM have not been detected in this library. In addition, the presence of SBEM mRNA in normal tissues has been investigated by three different laboratories.2, 3, 4 Using Northern dot blot analysis, SBEM mRNA expression was detected in normal breast and salivary glands but not in other normal tissues such as brain, ovary, uterus, prostate, uterus, and lung.3, 4 With a more sensitive microarray hybridization approach, Houghton et al2 showed that, besides salivary gland and breast, which expressed the highest level of SBEM mRNA, the only other normal tissues expressing detectable levels of this messenger were colon, kidney, and heart but not lung.2 On the other hand, other laboratories have reported the detection of MGB1 mRNA by reverse transcriptase-polymerase chain reaction (RT-PCR) in primary lung cancers.5, 6
Altogether, all database analyses and experimental data published have indicated that there is a higher likelihood of detecting MGB1 gene expression rather than SBEM in lung tissues. It was therefore surprising that the data presented by Koga et al showed SBEM to be highly expressed in normal lung tissues (more than 95%) and MGB1 to be undetectable in these samples.1
In order to clarify this issue, we analyzed by RT-PCR the synthesis of MGB1 and SBEM mRNAs in matched pairs of normal- and tumor-lung tissue samples collected from 15 patients. These patients with non-small-cell lung cancer (NSCLC) had undergone complete surgical resection of the lung tumor as their primary treatment (ie without prior radiotherapy or chemotherapy) between January 2002 and May 2003 at the ‘Hôpital Trousseau’ (Tours, France). As shown in Figure 1
, no SBEM expression was detectable in any of the 15 normal lung samples studied, while SBEM transcripts were present in three of 15 tumor samples (patients 7, 8, and 11). In addition, MGB1 transcripts were detected in 12 of 15 and 11 of 15 tumor and normal lung tissues, respectively.
|
Variations in the Cellular Composition of the Samples
Koga et al observed a difference in SBEM gene expression in the different cellular components of the normal lung, with a strong expression in the bronchial surface epithelium and the bronchial gland but no detectable expression in the peripheral parenchyme. Because of sampling irregularities, tissues samples may have included uneven proportion of ciliated epithelial cells, Kulchitsky cells, Clara cells, smooth muscle cells, pneumocyte type I and II, alveolar macrophages, or blood cells. This could affect the outcome of the analysis.
Pathology of the Tissue Samples
In addition to neoplastic lung tissue samples, normal lung tissue from an area distant to the tumor was identified and dissected by a lung pathologist. The presence of bronchioles and alveoli in normal tissue was further assessed by microscopy of paraffin section (data not shown). These samples may nevertheless differ from normal tissues selected in patients without lung cancer or other diseases. In contrast, Koga et al did not discuss the exact composition of their normal tissues.1
Contamination of Lung Samples by Skin Cells
Houghton et al2
found a detectable expression of SBEM by RT-PCR in skin. It is therefore possible that the detection of SBEM gene expression in some cases may have resulted from the introduction of skin cells during the process of samples retrieval.
In conclusion, most evidence to date has indicated that MGB1 is more likely to be expressed in normal lung than SBEM and does not support the Koga et al data. Further evaluations need to be carried out to confirm the status of SBEM gene expression in lung.
|
We thank Dr. De Muret (Département d’Anatomie Pathologique, CHRU de Tours, France) and Prof. Dumont (Département de Chirurgie Thoracique, CHRU de Tours, France) for providing the lung samples.
Footnotes
This work is supported by the Canadian Institute of Health Research, Friends You Can Count On, the ‘Institut pour la Recherche sur la Thrombose et l‘Hémostase’, and the ‘Ligue contre le Cancer’.2 F.H. is the recipient of a Manitoba Health Research Council award grant; J.R. of a grant from Governmental Research Departement; E.L. of a US Army Medical Research and Materiel Command career development award grant; and Y.M. has a Manitoba Medical Service Foundation career award.
F.H. and J.R. contributed equally to this work.
References
Aichi Cancer Center, Nagoya, Japan
Authors’ Reply:
We thank Hube et al for their interest in our article. In their analysis of 15 normal and 15 lung cancer tissues, they demonstrated that neither normal lung tissue nor lung cancers expressed mRNA for small breast epithelial mucin (SBEM), whereas the expression of mammaglobin A (MGB1) was frequently detected in those tissues. Those results contradict our findings.
From Northern dot blot hybridization analysis, SBEM was initially reported to be expressed in a restricted range of organs, ie breast and salivary glands.1 With more sensitive RT-PCR, the range of organs expressing this molecule was extended to include the colon, kidney, and heart.2 We demonstrated significant expression of SBEM in normal lung tissues,3 whereas Hube et al reported that SBEM was not expressed in lung tissue. A recent paper by Mitas et al further extended the range of expression to include the esophagus.4 Unfortunately, we excluded SBEM from further studies because it is not a definitive marker of breast cancer metastasis to the lung. Therefore, we cannot comment on further evidence of SBEM expression in normal lung tissues, in contrast to the expression of MGB1 (see below).
As indicated by Hube et al, it is possible that variation in the cellular composition of tissues may affect the expression status of SBEM. Our analysis with laser microdissection revealed that the main site of SBEM expression is the bronchial surface epithelium and bronchial glands, but not the peripheral parenchyma. However, when lung tissue is obtained, a mixture of bronchus and peripheral parenchyma is common. In our study, the positive expression of SBEM in 21 of 22 normal lungs suggested such a mixed cellular composition. Regarding the second comment, it is quite reasonable that tissue of normal appearance around a tumor may no longer be normal. To obviate this possibility, we included three samples of normal lung tissue that were resected due to metastatic tumors. All these normal lung tissues showed positive expression of SBEM. Regarding the third comment, contamination with skin tissue is unlikely, because the 22 normal lung tissues were obtained immediately after a lobectomy or partial resection at the pathology department, which was separated from the operating theater.
In terms of MGB1 expression, we further confirmed the organ-specific expression of MGB1. Paraffin sections of a total of 473 tumors from various organs were examined with immunohistochemistry. The primary tumors examined included 20 head and neck cancers, eight thyroid cancers, 244 breast cancers, 115 non-small-cell lung cancers, nine esophageal cancers, 19 gastric cancers, 16 colon cancers, three pancreatic cancers, 14 uterine cervical cancers (10 squamous cell carcinomas and four adenocarcinomas), nine endometrial cancers, and 13 ovarian tumors. The expression of MGB1 was restricted to breast and endometrial cancers, whereas none of the 115 lung cancers or 10 normal lung tissue samples were positive for MGB1 (manuscript in preparation). These results are quite consistent with previous work.5
6
7
Table 1
summarizes the data on MGB1 expression in normal lungs and lung cancers that have been reported to date, including in this communication. The restricted expression of MGB1 is also supported by other findings reported in the literature. Using expression profiling, Su et al reported a molecular classification based on gene subsets that could individually identify prostate, breast, lung, ovary, colorectal, kidney, liver, pancreatic, bladder, and gastroesophageal cancers.8
In addition to well-known classifiers such as TTF-1 for lung cancer, MUC-2 for colorectal cancer, and uroplakin II for bladder cancer, MGB1 was selected as a highly specific classifier of breast cancer. Bhattacharjee et al also demonstrated the molecular classification of lung adenocarcinomas with unsupervised hierarchical clustering based on expression profiles. In this analysis, a metastatic cancer from the breast was distinguished by nonlung signatures, including characteristic expression of the estrogen receptor and MGB1.9
These findings suggest that MGB1 expression in normal lung tissues is quite unlikely, in contrast to the finding of Hube et al. Further evaluation of this issue is required.
Acknowledgments
I thank Yutaka Hatanaka for excellent immunohistochemical staining of MGB1.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
