| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Letters to the Editor |
The Johns Hopkins University School of Medicine
It is with some puzzlement that we respond to the letter of Dr. Waisfisz et al. On review, we find no fault with our clinical study, and the letter appears not to challenge the validity of any of our specific data or statements. It accepts certain of our premises and evidence. It claims to find fault with some conclusions but with which particular conclusions it is not clear. The letter cites data that are not referenced and thus unavailable for our examination. Because of ambiguities regarding the underlying assumptions of certain statements made in the letter, a discursive "if ... then" response seems required and for which we may be at some disadvantage in attempting our full reply below. This is not optimal, for the authors make many points with which we and any experienced geneticist would readily agree. For example, in the case of some mutations, we specifically stated a lack of known ties to disease, and in a sample containing another variant, we established that the expression of the wild-type allele had been maintained. Indeed, for most missense mutations in the human genome, geneticists can generally assume that many or most will not be associated with disease. The difficulties attendant to the classification of germ-line missense mutations being well-known, the letter takes an opportunity to speculate on the functional classification of individual missense mutations in more detail than with which we might originally have felt comfortable, given the early stage of the related epidemiological and biochemical studies. We are aware of the many valuable contributions of Dr. Joenje and colleagues to the field of Fanconi research. We thus hesitate to put on the appearances of a strong and global disagreement if indeed ambiguities are clouding the exchange. Although we had made a request to the journal that additional clarification be obtained from the letter’s authors before our response, our request could not be entertained under the operative format. However, as would be the case for any scientific authors, we can readily embrace the not unattractive opportunity to provide some elaboration of our own. We thus thank Cancer Research for the forum, and herein, we endeavor to provide a discourse of value for the readers of the journal.
The letter offers a challenge, that "we think their conclusions are not necessarily supported by the data." There is, however, ambiguity as to what conclusions are being referred to because a particular published statement of ours is not specified. In the sentences immediately preceding this challenge, several potential conclusions are incorporated in the letter.
One interpretation as to the conclusion under challenge would be our manifest implication that in people not affected by FA1 (or by biallelic mutations of Fanconi genes), there exist in neoplasms pathogenic homozygous mutations of FANCC and FANCG that are of high interest to the cancer research community, in turn, implying that at least some of the mutations should be highly convincing, i.e., they should impair key functions of each gene. In the case of the FANCG gene, the authors of the letter have acceded as much, specifically citing our E105ter mutation in the cell line Hs766T, which is a homozygous mutation known to cause FA (1 , 2) . In the case of the FANCC gene, the authors specifically challenge two mutations in depth. As regards to the D195V missense mutation, we would have liked to have cited Dr. Joenje’s complementation study (3) had we been aware of it. This mutation remains of interest to researchers, however, in that the only two published observations were in a Fanconi patient (of unknown complementation group) and as a homozygous change in a pancreatic cancer (4 , 5) . The mild phenotype of the reported Fanconi patient has raised the possibility of a hypomorphic allele, which cannot be studied readily by standard complementation assays because of their engineered overexpression. Complementation studies are also inappropriate for any mutations that in the natural setting would produce a deficient level of protein expression, and levels of protein expression have not yet been studied in depth. As to whether the exon 14 frameshift mutation is relevant to a tumorigenic role, we find this particular mutation most convincing. It is thus important to examine the letter’s argument against the frameshift mutation. We are left at a disadvantage to address the data cited in the letter, "tags such as HA or GFP added to the FANCC COOH terminus are known to be without effect for the protein’s function," because the data are not referenced in the letter. We presume that the tagging of the ends of proteins that are both full-length and functional would be a methodology irrelevant to the issue here. We therefore do not find this argument in the letter to be convincing. Nevertheless, we can offer an independent commentary related to properties of the frameshift mutation of FANCC that we observed. Property 1: this is a mutation that would likely terminate protein translation prematurely. Most terminating and partial deletion mutations selected for during human tumorigenesis indeed affect message or protein stability and lead to a decrease in protein expression (such as seen with the well-studied mutations of the CDKN2A, RB1, and MADH4 genes in primary human tumors; Ref. 6, 7, 8 ). Thus, studies of overexpressed and reengineered proteins in an artificial model would often be expected to unintentionally correct the primary defect that had played a role in tumorigenesis. In this situation, studies of such an artificial model would be moot. Property 2: this is a truncation removing the COOH-terminal eight amino acids. It has been suspected, if not concluded (9) , that a major Fanconi-related function of the FANCC protein requires an intact COOH-terminal end. As shown by published studies of Dr. Joenje and others, missense and truncating mutations of FANCC, including those affecting the extreme COOH-terminal end and the final five amino acids, cause a severe Fanconi-associated phenotype in patients or in cellular assays (3 , 4 , 10, 11, 12, 13) , whereas a nonsense mutation of exon 1 is associated with a reinitiation of translation that preserves the COOH-terminal end and has a milder phenotype (14) . Property 3: this is a homozygous somatic mutation. Homozygous somatic truncating mutations in neoplasms exhibiting CIN are among the most convincing for the implication of a tumorigenic role (15 , 16) . The subclone attaining this FANCC frameshift mutation was able to out-compete innumerable clones having wild-type FANCC sequences to become the clone that killed the patient. If spurious sequence changes in CIN neoplasms were fairly common, one would not be able to make this type of inferential argument. The presence of passenger (spurious, chance) mutations, those not affecting function and not capable of providing a selective advantage to a subclone, would, however, be exposed by the common finding of somatic silent mutations in coding sequences (16) . Yet, during our years of studies of CIN pancreatic carcinomas comprising the sequence determination of 2–3 megabases of tumor DNA in search of somatic mutations, we have never observed a silent somatic mutation (17) . The rate of such mutations in small studies such as our recent publication must be very low to negligible. An estimate of the boundary on the upper frequency at which such nonsynonymous somatic passenger mutations are observed has been produced by a direct measurement in colorectal cancer, finding only three potential passenger somatic intragenic mutations in 3.2 megabases of sequenced tumor genome (18) . On the basis of these three known properties of the mutation in question, it seems beyond reason to conclude, as in the letter, that the exon 14 FANCC frameshift mutation did not affect gene function and contribute to tumorigenesis and death in this patient.
A second interpretation as to the conclusion under challenge could be our strong underlying premise that tumors with Fanconi deficits might be rational targets of mitomycin C or other interstrand DNA-cross-linking therapeutic agents. Although we find such a consideration self-evident, one must examine the possible assumptions or scenarios that would contradict this premise. Each assumption, however, would appear to encounter fatal disagreements with considerable bodies of published work. In the first and second scenario, it is possible that one could argue that BRCA2 is not a FA gene or that BRCA2 mutations do not affect persons with pancreatic cancer. If the assumption were the former, then the argument would conflict with the finding of homozygous or mixed heterozygous mutations of BRCA2 in cells typed as Fanconi complementation groups D1 and B (19) , conflict with studies of the BRCA2-deficient CAPAN1 cells (20) in which the chemosensitivity profile and other features closely match that of a Fanconi-like phenotype (21, 22, 23) , and conflict with a recent publication of Dr. Joenje that discussed BRCA2 as a Fanconi gene (24) . If the assumption were the latter as suggested strongly by the letter’s statement that "the suggestion that FA gene mutations play a significant role in the etiology of pancreatic cancer would seem premature," then the argument would conflict with clinical reports that 7–10% of apparently sporadic pancreatic cancers harbor homozygous BRCA2 mutations and that 12–17% of pancreatic cancer families have germ-line disease-inducing mutations of BRCA2 (these numbers exclude the exon 27 mutation; Refs. 20 , 25, 26, 27 ). It is possible that the argument rests upon a third assumption that the use of mitomycin C is not specifically supported by the known properties of cells of the Fanconi complementation groups and of BRCA2-deficient cells. If this is the assumption, then the argument would conflict with numerous studies of the chemical hypersensitivity of such cells (21, 22, 23 , 28) . It has been suggested by us and by others (28 , 29) that the major clinical rationale for a new use of mitomycin C, cisplatin, and potentially other interstrand DNA cross-linking agents could rest upon the determination of BRCA2 status, basing the suggestion on this chemosensitivity literature, on data showing that BRCA2 is the most commonly inactivated Fanconi gene in pancreatic and other cancers known to date, and on the imperfect and yet intriguing observances of prolonged clinical remissions of pancreatic cancer in some patients treated with mitomycin-containing protocols (30, 31, 32, 33) .
A third interpretation as to the conclusion under question is their concluding enumeration of "only one" and their skepticism as to whether "a significant proportion" of patients had mutations. This is not a phrase from our work. Estimated proportions being difficult to ascertain when dealing with small numbers, we provided a validated report that individual pancreatic cancer cases of onset <60 years and some cases <50 years had such changes. This distribution was compared with that of BRCA2 mutations, where a higher mean age had been found among our samples. These are the observed data, which were accurately represented in the title, abstract, and body of our publication. The frequencies in representative populations will be established in future studies of larger sample sets.
The only BRCA2-deficient human cancer cell line remains CAPAN1, a pancreatic cancer line in which mutation was reported by us in 1996 (20) . The only non-BRCA2 human cancer cell line with an undisputed Fanconi gene genetic inactivation published to date is Hs766T, a pancreatic cancer line in which mutation was reported by us recently (5) . The first BRCA2 mutation identified molecularly was a single pancreatic cancer having a BRCA2 homozygous mutation, reported by us in 1995 (34) . Perhaps the first plausible tie between Fanconi gene mutations and pancreatic cancer was reported as long ago as 1976; a consanguineous pedigree of a Scottish island was described in which a youth had FA, and the obligate mutation carriers had multiple instances of pancreatic and other cancers (35) . These seminal reports represent individual strings that can be used to assemble influential ropes of argument for the involvement of the FA pathway in pancreatic and other cancers. Although our recent studies are only an initiation of new investigations, we remain hopeful for a better treatment strategy in a specific subset of patients with pancreatic cancer: those with tumors defective in the Fanconi pathway.
FOOTNOTES
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 The abbreviations used are: FA, Fanconi anemia; CIN, chromosomal instability. 
Received 7/24/03. Accepted 7/25/03.
REFERENCES
-interferon in mice with a targeted disruption of the Fanconi anemia C gene. Blood, 88: 49-58, 1996.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
