Introduction
Chemoprevention only recently matured into a standard modality for controlling epithelial carcinogenesis. This maturity resulted from the landmark United States FDA3 approvals of tamoxifen for reducing breast cancer risk in three settings (1) and from FDA approvals of several agents for treating advanced premalignancy, or IEN, most recently the NSAIDs diclofenac and celecoxib in the IEN settings of actinic keratosis and FAP (2 , 3) . We were privileged to serve on the AACR Task Force on IEN that prepared the Special Article “Treatment and Prevention of Intraepithelial Neoplasia: An Important Target for Accelerated New Agent Development,” which appears elsewhere in this issue (3) . We now have the honor of writing “The Biology Behind… ” commentary on this seminal report documenting the importance of IEN end points to the field of cancer chemoprevention.
Although chemoprevention has come of age as a cancer-control modality (1, 2, 3, 4, 5) , the biology behind cancer chemoprevention, including treating and/or preventing IEN, has been tinged with riddle. One researcher touches the tail and claims, “It’s a snake of temporary inhibition.” Another touches the leg and declares, “It’s a tree of permanent reversal.” A third researcher touches the side and says, “It’s a wall of risk reduction.” As in the parable of the blind men and the elephant, most of us are blinded to one extent or another by our specialized perspectives on cancer chemoprevention. Although possibly encompassing permanent or complete prevention, the whole elephant certainly encompasses the delay of cancer development and all of the molecular biological complexity this implies.
This commentary will present strong evidence for the biological basis and clinical benefit of delay. We hope it may stimulate a new way of looking at cancer chemoprevention, especially for readers who believe that prevention must be tantamount to a complete or permanent incidence reduction, which we heartily dispute. We also will outline the novel concept of delaying cancer by molecular detours.
Chemoprevention and Biological Considerations of Cancer Delay
Cancer chemoprevention can be defined in practical clinical terms as the reduction in the rate of cancer development, or incidence, by single or combined agents for the period of a Phase III (cancer end point) trial (5) , as seen in the BCPT (1) . Animal carcinogenesis studies suggest broadening this definition by adding the concept of a rate reduction based on a fixed number of cancers developing over a longer time (in treated versus control animals). Regarding the first definition, it is not entirely clear whether the incidence reduction in the BCPT resulted from eradicating premalignant clones or delaying carcinogenic progression. It probably resulted from both. The second, animal-based definition, however, is the essence of cancer delay. Evidence from both the clinic (discussed in a later section) and the laboratory (discussed below in this section) indicates that a substantial part of chemopreventive activity involves delay.
Implying that the preventive effect lasts for a finite period (whether therapy is stopped or not), the concept of cancer or IEN delay is not new to chemoprevention (6, 7, 8, 9, 10) , having been addressed comprehensively in the laboratory with respect to the classic phases of chemical carcinogenesis. Delay terminology is used in describing the standard animal measures of chemopreventive efficacy, which include measures of decreases in the rate of tumor development, even if the incidence eventually returns to that in the control group, and of overall decreases in the incidence or number of tumors. The decreased rate of tumor development is measured by the increased time of tumor latency, which is the time between carcinogen exposure and either the first tumor or 50% of the overall tumor incidence. Active chemopreventive agents can produce a several-month shift to the right in the curves of the time-to-tumor-development (latency) and survival durations in rodents. Considering the large ratios of human:rodent tumor latencies and life spans (e.g., ∼50:1 for humans:mice), this shift could represent years or decades of increased time-to-cancer and survival in humans. “Even a delay in [bladder] tumor development [by retinoids] of as little as 7 to 10 weeks in the rat,” stated Hicks (and echoed by Moon et al.; Refs. 9 , 10 ), “could equate to an extra 5 or 6 years of symptom-free life for the human bladder-cancer patient.”
Sophisticated statistical models have been developed for analyzing animal studies with respect to the carcinogenic phases and measures of drug efficacy (11 , 12) . The ability of these models to assess delay is confounded by the usual termination of animal experiments before all of the possible latent tumors have developed in the treatment group and by forced carcinogenesis designed to produce a ∼100% cancer incidence in control animals in a short time.
For decades, researchers studying animals have accepted the potential clinical impact of cancer delay. Within the clinical research and practice community, however, considerations of cancer delay gained prominence only recently and did so because of the dramatic trial results and FDA approvals of tamoxifen in breast cancer prevention (1 , 5 , 13, 14, 15) . Needing many years or possibly decades of follow-up, it will be difficult, if not impossible, to prove cancer delay in the human primary prevention setting, which involves healthy, frequently high-risk, subjects (1) . Nevertheless, recent advances in our understanding of the molecular and cellular biology of carcinogenesis allow for highly educated reflections on cancer delay in the clinical setting.
The molecular basis of multistep carcinogenesis was first illustrated by Vogelstein et al. (16) in studies of human adenoma-carcinoma progression in the colon. The molecular biology of multistep carcinogenesis subsequently has been worked out in other epithelial sites, including the head and neck (Refs. 17, 18, 19, 20, 21 ; Fig. 1⇓ ), which exemplifies the molecular basis of cancer delay. Head and neck carcinogenesis involves accumulated molecular alterations that contribute to clonal expansion, intraepithelial spread, lesion heterogeneity, and drug resistance. First described 50 years ago as “field cancerization” in the setting of head and neck carcinogenesis (22) , multifocal carcinogenesis is another fundamental concept of chemoprevention and involves multiple genetically distinct clones and the lateral spread of genetically related preinvasive clones (4 , 20 , 21) .
Leukoplakia-carcinoma progression models in the oral cavity. Top panels, the clinical progression of a patient’s white oral leukoplakia lesion (left) to an oral cancer (right) that developed 3 years after complete leukoplakia resection. Middle panels, histological progression from hyperplasia to invasive cancer. Bottom panels, a molecular progression model, in which the accumulation of genetic alterations is more important than the order. The apparent loss of heterozygosity at 11q may represent allelic imbalance via cyclin D1 amplification at 11q13. Some molecular alterations, such as an autocrine growth loop involving transforming growth factor-α (TGF-α) and epidermal growth factor receptor (EGFR) overexpression, can occur in carcinogen-exposed histologically normal epithelium. Molecular alterations will differ depending on carcinogenic exposure [e.g., to cigarette smoke, betel nuts, or human papillomavirus (especially for oropharyngeal cancer)], and on genetic susceptibility (e.g., affected by certain glutathione S-transferase genotypes and by helicase defects involved with DNA-repair genes). FHIT, fragile histidine triad. Modified with permission of the Massachusetts Medical Society from a figure originally published in The New England Journal of Medicine (20) . Copyright © 2001 Massachusetts Medical Society. All rights reserved.
As indicated by advancing molecular studies, the lengthy process of carcinogenesis is multipath (23) as well as multistep and multifocal. These studies have provided new perspectives on chemoprevention and cancer delay. Depending on individual differences in gene-environment interactions [Refs. 20 , 24 ; e.g., involving DNA-repair capacity (25)] , an estimated 6-to-12 critical genetic events must accumulate, usually over 20 or more years, for clinically detectable cancer to develop after carcinogenic exposure. Slowing the mutational burden should delay IEN and cancer. Loeb suggested that the rate of accumulated mutations could be delayed by suppressing genomic instability and the mutator phenotype (26) . Therefore, agents producing a 2-fold reduction in random mutation rates early in carcinogenesis could double the latency period of cancer development (e.g., from 20 to 40 years). This concept is supported by DNA adduct studies of skin and liver carcinogenesis/chemoprevention (27 , 28) . Skin cancer delay in rodents was associated with slowing the mutation rate (27) . Molecular-targeting treatment of a DNA-repair defect reduced the rate of skin cancer in xeroderma pigmentosum patients (29) . The DNA helicases have been implicated in the mutator phenotype, genetic instability, and high cancer incidences, making them potentially important future molecular targets for chemoprevention by cancer delay (30) .
Fig. 2⇓ depicts a model of head and neck carcinogenesis illustrating our refined concept that cancer can be delayed by molecular detours, or by molecular-targeting approaches that detour multistep carcinogenesis down alternative pathways. The alterations of several genes, including p16 (on 9p), p53 (on 17p), and cyclin D1 (on 11q), are associated with the development of head and neck cancer (20 , 21) . Several agents may reverse the effects of these genetic alterations. Fig. 2⇓ illustrates this concept with a hypothetical detour at p53. p53 mutations are associated with decreased apoptosis (31) and carcinogenic progression (32) , and chemopreventive efficacy has been associated with enhancing apoptosis (33, 34, 35) . Gene therapy has shown promise in restoring wild-type p53, thereby increasing apoptosis in these cells (5 , 36) . Other potential molecular detours may involve correcting p16 inactivation with demethylating agents or gene therapy and suppressing cyclin D1 amplification or activity with antisense therapy or cyclin-dependent kinase inhibitors (20) . Combinations of molecular-targeting small molecules could enhance cancer delay by producing detours at multiple critical molecular events within the initiation, promotion, and progression phases of cancer development (1 , 5 , 20) .
Detouring critical genetic events to delay multistep carcinogenesis. Top panel, the multistep process of head and neck carcinogenesis in terms of the phenotypic, histological changes and the associated chromosomal sites (in the boxes) that undergo alterations. Various combinations of 6–12 genes are thought to be involved in critical alterations leading to oral cancer. Detour 2 (top panel) can occur relatively early and is associated with a relatively long delay in cancer development (bottom panel). Cascade of arrows emanating from detour 2, the complex of sub- and micropathways potentially involved in detouring carcinogenesis at any point during the multistep process (although not depicted, similar pathways would apply to detours 1, 3, and 4, top panel). Coming earlier in the process, it is likely that detour 1 would produce a greater delay (bottom panel). Coming later, detours 2–4 would produce lesser delays in cancer development (bottom panel), because carcinogenesis would have accumulated a greater burden of mutations and larger sheer numbers of progressing premalignant cells. Whereas single agents may produce the initial detours, it may be possible to design combinations of agents that work in multiple detour pathways to even further prolong latency. Bottom panel, X-axis, the years beginning with carcinogenic initiation; Y-axis, the total incidence of a particular cancer. Black curve at the left, the control (c, untreated multistep carcinogenesis); dotted green curve at the right, the hypothetically greatest delay possible, or an intervention before hyperplasia.
The preceding paragraph cites several molecular targets in the setting of head and neck cancer chemoprevention. Of course there are many other important molecular-targeting approaches with the potential to delay cancer in various settings, e.g., the induction of cell dormancy to undermine clonal outgrowth (such as with angiogenesis inhibitors) and apoptosis induction (such as with lipoxygenase modulators, cyclooxygenase inhibitors, and retinamides; Refs. 5 , 37, 38, 39, 40, 41, 42, 43, 44, 45 ). Recent molecular-targeting studies involving the APC pathway in colorectal carcinogenesis support the molecular detour concept. The effects of APC mutation, an early event in colorectal carcinogenesis, can be reversed by NSAID inhibition of the downstream (from APC)-target peroxisome proliferator-activated receptor δ (PPARδ; Ref. 44 ), which suggests a role for this detouring event in the established chemopreventive effects of NSAIDs (2 , 3 , 5 , 38) .
The only possibility of completely preventing cancer is to eliminate (e.g., by apoptosis) all of the initiated and more-advanced premalignant cells and the carcinogenic exposure (e.g., tobacco). Eliminating all premalignant cells without stopping carcinogenic exposure may reset the 20-year carcinogenic clock but will not completely prevent or eliminate cancer risk. This concept is supported by the surgical elimination of colorectal, oral, or breast IEN, which decreases cancer rates (3 , 20 , 46) . Definitive surgical resection, however, is not always possible or acceptable in prevention settings, e.g., in germ-line mutation carriers or in patients with multifocal diffuse premalignant lesions (20 , 47, 48, 49) . The potential of integrating chemoprevention with local therapy is indicated by the addition of tamoxifen to lumpectomy/radiation in treating DCIS, which reduced breast cancer and IEN events from 13.4% to 8.2% over a period of 5 years (versus lumpectomy/radiation alone; Ref. 50 ).
Aspects of Delay
Fig. 3⇓ depicts the three major aspects of cancer delay, reduced cancer rate followed by an increasing rate that (a) does not return to the baseline rate; (b) returns to the baseline rate; or (c) exceeds the baseline rate. The first two aspects could potentially be expected to result in clinical benefits of delayed cancer- and treatment-related morbidity and mortality and in reduced lifetime cancer incidence. The third aspect, with its potential to increase lifetime cancer incidence, would be an undesirable outcome of chemopreventive delay. There also are two potentially undesirable expressions of any of the three delay aspects: (a) enhancing the development of drug resistance, and (b) accelerating late (more aggressive) lesions while delaying early (more benign) lesions (51) . These negative effects are also potential concerns of delaying IEN progression. The potentially negative effects require careful long-term follow-up of positive chemoprevention trials, such as the landmark BCPT (1) and celecoxib/FAP trials (2 , 3 , 13) .
The three major aspects of cancer delay. Top graph (analogous to animal delay experiments), possible delay scenarios in human settings of virtually 100% cancer risk, such as in FAP or xeroderma pigmentosum patients. The three curves to the right, delayed cancer development followed by an increasing rate of development that does not return to the baseline (D1), that returns to the baseline (D2), or that exceeds the baseline rate (D3). D1, similar to data of DNA-repair-targeted delay in xeroderma pigmentosum patients; D3, similar to data of 13-cis-retinoic acid delay in xeroderma pigmentosum patients. Although not a 100% cancer-risk setting, the retinoid delay in head and neck cancer-related SPTs resembled the D2 curve. C, the control/baseline curve. Bottom graph, more typical scenarios of potential delay in major epithelial cancers, such as breast and prostate cancers (the curves are labeled similarly to the those in the top graph; exception: Y axis, annual incidence).
Clinical Models of Delay
This section presents three chemoprevention models as paradigms of clinical or molecular delay in cancer development. The use of retinoids in head and neck carcinogenesis is the best model of the molecular basis and provides the clinical proof-of-principle of cancer delay. The selective estrogen-receptor modulator-breast model best shows the preclinical evidence and potential clinical benefits of cancer delay. NSAIDs in colorectal carcinogenesis are an excellent model to show the clinical benefits of delaying IENs.
Retinoid-Head and Neck Model and Cancer Delay.
Several trials involving retinoids and interferon α in head and neck IEN suggest that, without a molecular complete response, cancer is being delayed rather than completely prevented (52, 53, 54, 55, 56, 57, 58) . Single-agent 13-cis-retinoic acid produced phenotypic responses but did not eliminate clones with p53 mutations. Prolonged 13-cis-retinoic acid therapy delayed IEN progression, although lesions eventually became resistant to the retinoid; and lesion progression, or drug resistance, was associated with persistent genotypic alterations and cancer development. These data suggest that cancer delay could result from prolonging the interval between critical genetic events in the multistep process. Combined retinoic acid-interferon therapy eradicated some clones; others persisted, despite complete phenotypic reversion, and probably retained the capacity to progress into phenotypically apparent lesions (20 , 21 , 57, 58, 59) . The combination results suggest that eradicating premalignant clones also can achieve cancer delay.
The effects of 13-cis-retinoic acid on SPTs associated with head and neck cancer (60) provided the proof-of-principle of cancer-delay’s clinical benefit. One year of high-dose 13-cis-retinoic acid produced a significant annual reduction in the incidence of SPTs, which disappeared 3 years after stopping treatment (61) . Although lasting only 3 years, this delay in SPT development was still clinically important, because SPTs are the leading cause of cancer death in early-stage-head-and-neck-cancer patients (62) .
The relatively short 3-year delay in the head and neck SPT setting raises the issue of the timing of a preventive intervention. The three general settings of prevention are: primary (healthy, frequently high-risk); secondary (premalignancy); and tertiary (treated cancer patients with no present signs of disease; Ref. 24 ). Because tertiary prevention confronts very advanced stages of fieldwide premalignancy (63) and likely-remaining microscopic cancer after definitive primary cancer therapy (64) , it should produce lesser delays than may occur with earlier interventions in the primary or secondary settings.
Selective Estrogen-Receptor Modulator-Breast Model and Cancer Delay.
Selective estrogen-receptor modulators can prolong the latency of breast tumors in animals by weeks or months. The animal evidence suggests that tamoxifen’s protective effects wear off after stopping treatment and that prolonged tamoxifen treatment can prolong the cancer delay. Jordan showed that tamoxifen treatment for 1 month beginning 1 month after 7,12-dimethylbenz[a]anthracene (DMBA)-induced carcinogenesis in rats produced a dose-related delay in palpable mammary tumor development (65) . However, tumors appeared in all of the treated animals after tamoxifen was stopped, although overall tumor burden was reduced. Prolonged tamoxifen treatment (e.g., for 6 months) prevented tumors in over 90% of the animals during treatment. How long the preventive delay lasts in humans is undefined, however. Five or 10 years of tamoxifen appear to reduce SPT incidence (to an equivalent degree) for at least 10 years (66) . Recent data in patients with DCIS [National Surgical Adjuvant Breast and Bowel Project (NSABP) B24 trial] indicate that, after 5 years of tamoxifen treatment, the reduction in breast-cancer-incidence reduction is persisting for 7 years. The preventive value of longer-term or continuous tamoxifen likely never will be tested in humans, at least not in the primary prevention setting (healthy, frequently high-risk; Ref. 24 ), because of this agent’s serious potential side effects (1) .
The use of tamoxifen in patients at high risk of developing breast-cancer is a paradigm of the clinical benefits of cancer delay. Tested in >50,000 women, tamoxifen has produced remarkably consistent invasive-cancer prevention/risk reduction results in the narrow range of 43–49% and is FDA-approved in the three distinct settings of: healthy high-risk women (primary prevention); the IEN DCIS (secondary prevention); and early-stage breast cancer (tertiary prevention; Refs. 1 , 24 ). The studies in the first two settings also found similar reductions in the development of IEN [lobular carcinoma in situ (LCIS), DCIS]. Estimates based on the BCPT (a definitive breast cancer prevention trial in 13,388 women) suggest that tamoxifen could prevent up to 500,000 primary breast cancers in the United States over a 5-year period. Another important potential benefit of tamoxifen in delay of breast cancer, depending on the reliability and development of risk monitoring tools (e.g., ductal lavage; 67 ), would be to delay prophylactic bilateral mastectomy in younger women at very high-risk because of inherited BRCA2 mutations (68 , 69) .
The BCPT was largely responsible for raising the debate on the value of cancer delay in the clinical chemoprevention setting (14 , 15) , a debate taken up during the September 1998 Oncologic Drugs Advisory Committee (ODAC) meeting that considered the approval of tamoxifen for the BCPT setting of healthy, high-risk women. For the most part, chemoprevention does not conform neatly to the widespread expectation that preventing or protecting against a disease entails taking a pill or shot to achieve virtually 100% protection, as with many vaccines. Phase III prevention trials generally target reductions in cancer incidence or risk of from 25% to 50%. Although these reductions are not complete prevention, they certainly produce clinical benefits by reducing the catastrophic effects of breast, colon, and other major epithelial cancers (benefits that are countered to lesser or greater degrees by expected and unexpected adverse effects). Chemoprevention in this regard parallels cancer therapy, in which the word “treatment” (implying the ability to reduce cancer burden) generally replaces “cure” with respect to major epithelial cancers. It may be possible, however, to use vaccines or antimicrobial agents to block the initiation of, or completely prevent, certain cancers caused by infectious diseases, such as cervical cancer associated with human papillomavirus, gastric cancer associated with Helicobacter pylori, and hepatocellular carcinoma associated with hepatitis B and C viruses (5 , 70 , 71) .
A BCPT-related criticism of cancer delay is that it may not decrease cancer mortality because only early lesions may be delayed (14 , 15) . Even a delay in less aggressive lesions, however, could have a major impact on morbidity and quality of life. Another BCPT criticism is that it delayed clinical cancer via treating subclinical, microscopic invasive cancer (1 , 5 , 14) . Extensive statistical modeling on the BCPT data suggests that some of the BCPT reduction in breast cancer incidence resulted from the effects of tamoxifen on microscopic invasive cancer (72) . Nevertheless, delaying the expression of clinical disease by treating microscopic invasive cancer (or microscopic DCIS) would be a positive effect of delay (73) . Other modeling on BCPT data suggests that tamoxifen may prolong overall and quality-adjusted survival in the highest-risk subjects (74) .
NSAID-Colorectal Model and Cancer Delay.
The NSAID-colorectal carcinogenesis model illustrates the potential clinical benefits of IEN (adenoma) delay. As defined in the AACR IEN Task Force article in this issue of Clinical Cancer Research (3) , IEN is a noninvasive lesion that has genetic abnormalities, loss of cellular control functions, and some phenotypic characteristics of invasive cancer and that is associated with a substantial cancer risk. These and other IEN characteristics, including intraepithelial spread, heterogeneity, and drug resistance (3 , 20) , make IEN difficult to eradicate with chemoprevention. Just delaying IEN, however, could be clinically beneficial, even without complete and permanent cure or cancer delay, by reducing or delaying IEN and its screening-and-therapy-related morbidity. It also is possible that IEN delay would improve cancer-related morbidity and mortality. Observational studies have found decreased incidence of adenoma in conjunction with decreased incidence or mortality of colorectal cancer associated with NSAIDs (3) . The incidence of sporadic adenomas has been reduced in recent randomized clinical trials of NSAIDs. Whether delaying cancer or just the premalignancy, IEN delay could extend the years of healthy, high-quality life. As described earlier in the context of the retinoid-head and neck IEN model (20 , 56, 57, 58) , it is important to evaluate IENs in clinical chemoprevention trials, when possible or feasible, for genetic and molecular progression. The clinical benefit of partial IEN response to treatment may depend on the molecular profile of the residual IEN (compared with that of IEN in placebo-treated subjects). A potential drawback of IEN delay was illustrated by studies of the NSAID, sulindac, in FAP (75 , 76) . Adenomatous polyps reappearing after sulindac suppression, or delay, were more difficult to screen because of a flattened appearance (in contrast to the situation with tamoxifen, which may enhance mammographic screening by its antiestrogenic effect of decreasing breast density).
Celecoxib (a selective inhibitor of cyclooxygenase-2) has been FDA approved for reducing the IEN burden of adenomatous polyps in FAP patients (2) , who have an inherited germ-line APC mutation causing hundreds of polyps by the teen years and conferring virtually a 100% risk of colon cancer by age 50. The sulindac-FAP studies suggested that NSAIDs delay, rather than permanently arrest, polyp development (77) . NSAID delay of adenomatous polyps could delay prophylactic proctocolectomy in children until they can complete high school and cope better with the morbidity of this procedure. High-dose celecoxib significantly reduced the number of adenomatous polyps in patients by 28% (compared with placebo). Combinations of NSAIDs and other drugs based on molecular interactions (43 , 78) may improve this effect in FAP, and agent combinations may be required generally for delay in the settings of late-stage IEN or SPT risk (79) , which involve the accumulation of an advanced number of molecular abnormalities.
Conclusions
The ideal, and possibly general perception of the role, of cancer chemoprevention is to develop vaccine-like regimens capable of eradicating cancer risk. After decades of extremely hard-won incremental advances in reducing epithelial cancer incidence and risk, researchers now realize that chemopreventive agents rarely, if ever, will completely or permanently prevent the enormously complex progression of premalignancy to malignancy in many sites. A major component of chemopreventive success will be measured by periods of delayed cancer development, morbidity, and mortality.
The duration of the delay could depend on how early the intervention begins within the multistep premalignancy process and on risk responses determined by individual carcinogenic exposures and genetic susceptibilities. Treating premalignancy in its earliest stages could extend the latency period by prolonging the intervals between more of the genetic events along the cancer development pathway. Although tamoxifen did not decrease the development of estrogen-receptor-negative cancer in the high-risk subjects of the BCPT (13) , animal studies have suggested that selective estrogen-receptor modulators may delay estrogen-receptor-negative breast tumor formation if given early enough in the carcinogenic process (73 , 80) . As suggested by the use of retinoids in head and neck SPT prevention, treating premalignancy during later stages may produce smaller gains in delay, because later stages involve an advanced number of molecular alterations, and fewer genetic events and intervals remain to be prolonged. Retinoids achieved even smaller SPT delays in skin cancer patients with xeroderma pigmentosum (81) , which has DNA helicase defects and other factors that worsen gene-environment interactions and genomic instability and increase the mutational burden. Intervening in earlier preinvasive stages may also limit multifocal spread. Removing the carcinogenic exposure (e.g., by smoking cessation) should enhance chemopreventive delay, whether earlier or later in the process.
The list of more common terms describing or defining chemopreventive activity includes “suppression,” “arrest,” “reversal,” “block,” or “inhibition” of premalignancy, and more recently, “cancer risk or incidence reduction.” “Delay” should be added to this list and removed from any list of pejorative terms used to discount chemoprevention. A pejorative and, we believe, mistaken view of delay came to the fore when critics discounted the BCPT’s clinical benefit because they thought, in part, that tamoxifen may have merely delayed cancer development (15) . Perhaps Sporn’s classic definition of chemoprevention, “to arrest or reverse premalignant cells” [Sporn et al. (7) and Sporn and Newton (82)] connotes permanent or absolute prevention to many people, which helps to overshadow the chemopreventive value of delay.
On the basis of the strong evidence from animal studies and suggestive evidence from clinical/translational studies, it is likely that clinical cancer chemoprevention is virtually synonymous with cancer delay (with a few exceptions, such as vaccination against diseases that cause cancer). Because the median age of cancer diagnosis in the United States is 70 years and average life expectancies are 74 years for men and 79 years for women, cancer delay can be tantamount to permanent prevention for many people. This clinical interpretation of the field was presaged by Wattenberg in 1966 (6) , when he wrote, “Even partial additional protection… which results in a prolongation of the latent period, might have the effect of preventing or minimizing significant manifestations of a malignant process during a normal life span,” and by Sporn and Newton in 1979 (82) , who wrote “Phenotypic suppression is not the ultimately desirable way to prevent cancer… . [but is] a practical approach to prevention by extending the latency period.” Cancer delay and the resulting clinical benefits will stand in the front lines of cancer chemoprevention until permanent or absolute prevention can be achieved.
Footnotes
-
↵1 Supported in part by the Cancer Center Support Grant CA16672 from the National Cancer Institute, NIH. S. M. L. holds the Anderson Clinical Faculty Chair for Cancer Treatment and Research. W. K. H. is an American Cancer Society Clinical Research Professor and Fellow National Foundation Cancer Research and holds the Charles A. LeMaistre Distinguished Chair in Thoracic Oncology, M. D. Anderson Cancer Center.
-
↵2 To whom requests for reprints should be addressed, at Department of Thoracic/Head & Neck Medical Oncology, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Box 80, Houston, TX 77030. Email: whong{at}mdanderson.org
-
↵3 The abbreviations used are: FDA, Food and Drug Administration; IEN, intraepithelial neoplasia; NSAID, nonsteroidal anti-inflammatory drug; FAP, familial adenomatous polyposis; BCPT, Breast Cancer Prevention Trial (as reported in Ref. 13 ); APC, adenomatous polyposis coli; DCIS, ductal carcinoma in situ; SPT, second primary tumor.
- Received December 21, 2001.
- Revision received December 28, 2001.
- Accepted December 31, 2001.