Migration rules: tumours are conglomerates of self-metastases.

Tumours are heterogeneous populations composed of different cells types: stem cells with the capacity for self-renewal and more differentiated cells lacking such ability. The overall growth behaviour of a developing neoplasm is determined largely by the combined kinetic interactions of these cells. By tracking the fate of individual cancer cells using agent-based methods in silico, we apply basic rules for cell proliferation, migration and cell death to show how these kinetic parameters interact to control, and perhaps dictate defining spatial and temporal tumour growth dynamics in tumour development. When the migration rate is small, a single cancer stem cell can only generate a small, self-limited clone because of the finite life span of progeny and spatial constraints. By contrast, a high migration rate can break this equilibrium, seeding new clones at sites outside the expanse of older clones. In this manner, the tumour continually 'self-metastasises'. Counterintuitively, when the proliferation capacity is low and the rate of cell death is high, tumour growth is accelerated because of the freeing up of space for self-metastatic expansion. Changes to proliferation and cell death that increase the rate at which cells migrate benefit tumour growth as a whole. The dominating influence of migration on tumour growth leads to unexpected dependencies of tumour growth on proliferation capacity and cell death. These dependencies stand to inform standard therapeutic approaches, which anticipate a positive response to cell killing and mitotic arrest.

Radiation-produced chromosome aberrations: colourful clues.

Ionizing radiation produces many chromosome aberrations. A rich variety of aberration types can now be seen with the technique of chromosome painting. Apart from being important in medicine and public health, radiation-produced aberrations act as colorful molecular clues to damage-processing mechanisms and, because juxtaposition of different parts of the genome is involved, to interphase nuclear organization. Recent studies using chromosome painting have helped to identify DNA double-strand-break repair and misrepair pathways, to determine the extent of chromosome territories and motions, and to characterize different aberration patterns left behind by different kinds of radiation.

Her-2-neu expression and progression toward androgen independence in human prostate cancer.

BACKGROUND: Human prostate cancers are initially androgen dependent but ultimately become androgen independent. Overexpression of the Her-2-neu receptor tyrosine kinase has been associated with the progression to androgen independence in prostate cancer cells. We examined the expression of Her-2-neu in normal and cancerous prostate tissues to assess its role in the progression to androgen independence. METHODS: Prostate cancer tissue sections were obtained from 67 patients treated by surgery alone (UNT tumors), 34 patients treated with total androgen ablation therapy before surgery (TAA tumors), and 18 patients in whom total androgen ablation therapy failed and who developed bone metastases (androgen-independent [AI] disease). The sections were immunostained for Her-2-neu, androgen receptor (AR), prostate-specific antigen (PSA), and Ki-67 (a marker of cell proliferation) protein expression. Messenger RNA (mRNA) levels and gene amplification of Her-2-neu were examined by RNA in situ hybridization and fluorescent in situ hybridization(FISH), respectively, in a subset of 27 tumors (nine UNT, 11 TAA, and seven AI). All statistical tests were two-sided. RESULTS: Her-2-neu protein expression was statistically significantly higher in TAA tumors than in UNT tumors with the use of two different scoring methods (P =.008 and P =.002). The proportion of Her-2-neu-positive tumors increased from the UNT group (17 of 67) to the TAA group (20 of 34) to the AI group (14 of 18) (P<.001). When compared with UNT tumors, tumor cell proliferation was higher in AI tumors (P =.014) and lower in TAA tumors (P<.001). All tumors expressed AR and PSA proteins. Although Her-2-neu mRNA expression was high in TAA and AI tumors, no Her-2-neu gene amplification was detected by FISH in any of the tumor types. CONCLUSIONS: Her-2-neu expression appears to increase with progression to androgen independence. Thus, therapeutic targeting of this tyrosine kinase in prostate cancer may be warranted.

Measurement of potential doubling time for human tumor xenografts using the cytokinesis-block method.

Estimates of the potential doubling time (Tpot) of seven human tumor xenografts were made using a cytokinesis-block method. This method is currently being investigated as an alternative to flow cytometric assays using the administration of a thymidine analog in the measurement of Tpot. If perfected, the cytokinesis-block method of measuring Tpot would be advantageous as a predictive assay, in that no label is administered to the tumors in situ. Xenografts were grown in nude mice, and following tumor excision and disaggregation, tumor cells were cultured with the cytokinesis-blocking agent cytochalasin B. The flux of cells through mitosis was marked by the accumulation of multinucleate cells. By counting the total number of nuclei as a function of time, the effective population growth was observed. Tpot values were obtained by fitting suitable exponential least squares curves to the data, with the doubling time indicated by the fitted functions. For the seven tumors studied, a significant spread in growth rates was observed. Tpot values generated by this method ranged from approximately 2 days for a rapidly growing squamous cell carcinoma of the pharynx, FaDu, to approximately 7.5 days for the slower growing glioblastoma multiforme U251-MG. These values are compared with standard 5-iododeoxyuridine Tpot measures and volume doubling times obtained by Perez et al. (Cancer Res., 55: 392-398, 1995) for the same tumor xenografts. Although individual Tpot values varied between these methods, the ranking of the seven tumors in order of Tpot times was the same regardless of method. In addition to estimating Tpot, for each of the tumors, the fraction of clonogenically dead cells that was microscopically apparent, including apoptotic cells and cells expressing micronuclei, was determined as a function of time in culture. Tracking this in vitro cell loss rate provides information on the adjustment of these primary tumor cells to in vitro culture, a factor that needs to be addressed when determining how in vitro measurements of Tpot can be effectively related to in vivo measurements.

Mammary fibroblasts may influence breast tumor angiogenesis via hypoxia-induced vascular endothelial growth factor up-regulation and protein expression.

Recent studies demonstrate the relationship of microvessel density to malignant progression in breast cancer (N. Weidner, J. P. Semple, W. R. Welch, and J. Folkman, N. Engl. J. Med., 324: 1-8, 1991), underscoring the importance of angiogenesis in this tumor. Crucial in tumor angiogenesis are the paracrine actions of tumor-secreted factors (e.g., vascular endothelial growth factor), which have been thought to derive from the tumor epithelial cells themselves. We demonstrate that in response to hypoxic conditions, human mammary fibroblasts dramatically up-regulate vascular endothelial growth factor mRNA and increase vascular endothelial growth factor protein levels in accordance with the degree of oxygen deprivation. Thus, mammary stromal cells, only recently considered in the regulation of breast carcinomas, may play a hitherto unrealized role in breast cancer angiogenesis.

Tumor development under angiogenic signaling: a dynamical theory of tumor growth, treatment response, and postvascular dormancy.

The effects of the angiogenic inhibitors endostatin, angiostatin, and TNP-470 on tumor growth dynamics are experimentally and theoretically investigated. On the basis of the data, we pose a quantitative theory for tumor growth under angiogenic stimulator/inhibitor control that is both explanatory and clinically implementable. Our analysis offers a ranking of the relative effectiveness of these inhibitors. Additionally, it reveals the existence of an ultimate limitation to tumor size under angiogenic control, where opposing angiogenic stimuli come into dynamic balance, which can be modulated by antiangiogenic therapy. The competitive influences of angiogenically driven growth and inhibition underlying this framework may have ramifications for tissue size regulation in general.

Detection of an intrinsic marker in hypoxic cells.

An autoradiographic method is presented for detecting, within a cell population, those cells which have been subjected to chronic hypoxia. No radioisotope is administered; rather the photographic emulsion is chemically reduced by intrinsic constituents of the cells. Hypoxic regions in the sandwich system, a multicellular in vitro tumor model, were detected in this manner. These regions were then compared with hypoxic sandwich regions as demonstrated by [3H]misonidazole labeling. Auxiliary studies, including studies on hypoxic monolayers, were consistent with the sandwich results. In all cases, the intracellular distribution of the chemographic grains was found to be cytosolic. Often the grains were clustered near the nucleus, perhaps in the region of the endoplasmic reticulum and the Golgi. We conclude that cells in a state of hypoxia and nutrient deprivation similar to that found in solid tumors retain a detectably altered biology for a significant period after reoxygenation. Therefore systematic methods of detecting previous hypoxia in histological tumor sections are feasible.

Adeno-associated virus-mediated delivery of antiangiogenic factors as an antitumor strategy.

Antiangiogenic tumor therapies have recently attracted intense interest for their broad-spectrum action, low toxicity, and, in the case of direct endothelial targeting, an absence of drug resistance. To promote tumor regression and to maintain dormancy, antiangiogenic agents need to be chronically administered. Gene therapy offers a potential way to achieve sustained therapeutic release of potent antiangiogenic substances. As a step toward this goal, we have generated recombinant adeno-associated virus (rAAV) vectors that carry genes coding for angiostatin, endostatin, and an antisense mRNA species against vascular endothelial growth factor (VEGF). These rAAVs efficiently transduced three human tumor cell lines tested. Transduction with an rAAV-encoding antisense VEGF mRNA inhibited the production of endogenous tumor cell VEGF. Conditioned media from cells transduced with this rAAV or with rAAV-expressing endostatin or angiostatin inhibited capillary endothelial cell proliferation in vitro. Antiangiogenic rAAVs may offer a novel gene therapy approach to undermining tumor neovascularization and cancer progression.

Full-color painting reveals an excess of radiation-induced dicentrics involving homologous chromosomes.

PURPOSE: To determine the ratio of homologous to heterologous dicentric chromosomes induced in human cells by ionizing radiation. This ratio is influenced by, and thus potentially informative about, underlying DNA damage/repair/misrepair processes and also the geometry of individual chromosome domains within the interphase nucleus. MATERIALS AND METHODS: 24-color mFISH (multiplex fluorescent in situ hybridization) was used to determine the ratio of 1-color (homologous) to 2-color (heterologous) dicentrics produced in human lymphocytes or fibroblasts by gamma-rays, alpha particles, or iron ions at various doses. Assuming that randomness independent of homology holds, the expected homologue:heterologue ratio for diploid human male cells is approximately 0.024, as shown by deriving a formula applicable to simple interchanges and then extending the result, via Monte Carlo simulation, to the general situation where complex aberrations are also considered. RESULTS AND CONCLUSIONS: There was a substantial excess of homologous dicentrics, with probability of occurrence by chance less than 0.02 for each of the three radiations and only about 10(-8) for all the data combined. Overall, approximately 18 homologous dicentrics were expected but 47 were found, including 11 involving chromosome 1. Observed excesses were similar for both sparsely and densely ionizing radiations. Geometric proximity of homologues is a possible explanation for the overabundance; in that case more extensive statistics should eventually uncover a linear energy transfer (LET) dependence. An alternative possibility, not ruled out by the present data, is homology-dependent misrepair.

Biological effects of proton radiation: an update.

Proton radiation provides significant dosimetric advantages when compared with gamma radiation due to its superior energy deposition characteristics. Although the physical aspects of proton radiobiology are well understood, biological and clinical endpoints are understudied. The current practice to assume the relative biological effectiveness of low linear energy transfer (LET) protons to be a generic value of about 1.1 relative to photons likely obscures important unrecognised differentials in biological response between these radiation qualities. A deeper understanding of the biological properties induced by proton radiation would have both radiobiological and clinical impact. This article briefly points to some of the literature pertinent to the effects of protons on tissue-level processes that modify disease progression, such as angiogenesis, cell invasion and cancer metastasis. Recent findings hint that proton radiation may, in addition to offering improved radio-therapeutic targeting, be a means to provide a new dimension for increasing therapeutic benefits for patients by manipulating these tissue-level processes.

Comparison of 3H-misonidazole binding between CHO and 9L cells using the sandwich system.

3H-misonidazole binding of 9L cells was compared with that of CHO cells using an in vitro tumor analog, the sandwich system. In sandwiches there is a gradient of microenvironments, with cells adjacent to the necrotic center subjected to low concentrations of oxygen and glucose and to high concentrations of metabolites. Mixed sandwiches, having 9L and CHO cells interspersed, were used along with sandwiches of each individual cell line. MISO binding was assessed in situ, using autoradiography. Grains per cell were counted and detailed statistics were obtained on the variation in MISO binding among cells located in the same microenvironment. In all cases binding in the regions near the necrotic center was more than 50 times the binding found at the sandwich edge and found in control monolayers, indicating radiobiological hypoxia near the necrotic center. 9L cells began to significantly increase binding of MISO metabolites at a somewhat higher oxygen concentration than did CHO cells. At all oxygen tensions, average per cell binding of the 9L cells was 3 times or more that of the CHO cells, a factor greater than can be explained by the ratio of cell volumes alone. Statistical analyses of the variation in binding among cells in a given microenvironment give some evidence that in the mixed CHO/9L sandwiches there are interactions between the cells of the two different lines which affect the growth patterns of the cells. No preferred binding of misonidazole in the nucleus or cytoplasm was noted within the cells.

3H-misonidazole labeling and viability of hypoxic cells in the sandwich system, an in vitro tumor analogue.

3H-misonidazole was used as a marker of hypoxic cells in an in vitro tumor analogue, the sandwich system. MISO binding was assessed in situ, using autoradiography. Binding profiles indicate that there are regions of radiobiological hypoxia surrounding the necrotic center in sandwiches of the V79 cell line and in sandwiches of the 9L cell line. Grains per cell were counted and detailed statistics on the variation of intrinsic binding among cells in the same microenvironment are presented. There is a systematic decrease in the standard deviation of grains per cell as one examines populations of cells further and further from the nutrient and oxygen source. Kinetic studies show that the growth fraction of the cell population also decreases with distance from the nutrient source. These findings taken together suggest that MISO binding is proportional to cell size and cells in the inner noncycling portion of the sandwich are more nearly uniform in size. Sandwich cells which exhibit heavy MISO binding, and are presumably radiobiologically hypoxic, were shown to be still viable if restored to good nutrient and oxygen conditions.

A convenient extension of the linear-quadratic model to include redistribution and reoxygenation.

PURPOSE: At present, the linear-quadratic model for cellular response to radiation can incorporate sublethal damage repair and repopulation. We suggest an extension, termed LQR, to include also the other two "Rs" of radiobiology, cell cycle redistribution, and reoxygenation. METHODS AND MATERIALS: In this approach, redistribution and reoxygenation are both regarded as aspects of a single phenomenon, which we term resensitization. After the first portion of a radiation exposure has decreased the average radiosensitivity of a diverse cell population by preferentially sparing less sensitive cells, resensitization gradually restores the average sensitivity of the population towards its previous value. The proposed LQR formula is of the same form as the original LQ formula, but with two extra parameters, an overall resensitization magnitude and a characteristic resensitization time. The LQR model assumes that resensitization is monotonic rather than oscillatory in time, i.e., always tends to increase average cellular sensitivity as overall time increases. We argue that this monotonicity assumption is likely to hold in clinical situations, though a possible extension is discussed to account for oscillatory decay of resensitization effects. RESULTS: The LQR model gives reasonable fits to relevant experimental data in the literature, reproducing an initial rise in cell survival, due to repair, as the treatment time is increased, followed by a resensitization-related decrease in survival due to redistribution and/or reoxygenation for treatment times of the order of the cell cycle time, and a final survival increase due to repopulation as the treatment time is increased still further. CONCLUSION: The LQR model is a simple and potentially useful extension of the LQ model for computing more realistic isoeffect relations for early responding tissues, including tumors, when comparing different radiotherapeutic protocols.

Distinct mathematical behavior of apoptotic versus non-apoptotic tumor cell death.

PURPOSE: The presence or absence of a p53-dependent apoptosis response has previously been shown to greatly influence radiosensitivity in tumor cells. Here, we examine clonogenic survival curves for two genetically related oncogene transformed cell lines differing in the presence or absence of p53 and apoptosis. Solid tumor radiosensitivity patterns have been previously described for these lines. MATERIALS AND METHODS: Oncogene-transformed fibroblasts derived from E1A + Ras transfection of p53-wild-type or p53-null mouse embryonic fibroblasts were plated as single cells and irradiated at increasing radiation doses in single fractions from 1.5 to 11 Gy. Clonogenic cell survival assays were obtained. Survival data are fit to a linear-quadratic relationship: S = e(-alphaD-betaD2). Apoptosis was assessed and quantitated morphologically by staining with the fluorescent nuclear dye DAPI, by TUNEL assay for DNA fragmentation, and by measurement of apoptotic cysteine protease cleavage activity in cytosolic extracts. RESULTS: Whereas radiation triggers massive apoptosis in the presence of p53, it produces no measurable DNA fragmentation, apoptotic cysteine protease cleavage activity, or morphological changes of apoptosis in the cells lacking p53. These contrasting mechanisms of death display dramatically different quantitative behavior: log-survival of apoptotic cells is linearly proportional to dose (S = e(-alphaD)), whereas survival of non-apoptotic (p53 null) is linear-quadratic with a significant quadratic contribution. The surviving fraction at 2 Gy (SF-2) for p53-null cells was 70% verses 12% for p53-intact cells. CONCLUSIONS: In this system, apoptosis appears to exhibit a dominance of single-event which produces a very high alpha/beta ratio, and no significant shoulder; whereas non-apoptotic death in this system exhibits a comparatively small linear component, a low alpha/beta ratio, and a larger shoulder.

Quantitative analysis of radiation-induced chromosome aberrations.

We review chromosome aberration modeling and its applications, especially to biodosimetry and to characterizing chromosome geometry. Standard results on aberration formation pathways, randomness, dose-response, proximity effects, transmissibility, kinetics, and relations to other radiobiological endpoints are summarized. We also outline recent work on graph-theoretical descriptions of aberrations, Monte-Carlo computer simulations of aberration spectra, software for quantifying aberration complexity, and systematic links of apparently incomplete with complete or truly incomplete aberrations.

Cell resensitization during protracted dosing of heterogeneous cell populations.

When a tumor or other heterogeneous cell population is acutely exposed to ionizing radiation (or, for that matter, to chemotherapeutic agents or hyperthermia), cells that happen to be more sensitive will be preferentially removed, leaving behind a population more resistant as a whole. However, under broadly applicable assumptions, we here demonstrate mathematically that there is a natural tendency of the postirradiation population to recover from the irradiation in such a manner as to restore its original sensitivity composition, i.e. to undergo "resensitization". An important consequence in radiotherapy is that, if a fixed total radiation dose is delivered in a more protracted manner, e.g. as several fractions or as a continuous dose at low dose rate, resensitization occurring over the course of dose delivery will result in greater cell killing than would otherwise have occurred. That is, for a cell population with any form of diversity in radiosensitivity, the influence of redistribution is to make any prolonged dose more damaging than an acute dose of the same magnitude. This tendency toward an "inverse dose-rate effect" may be masked in practice by countervailing effects, such as repair of sublethal damage, but the tendency is demonstrated to hold under very general circumstances, being a consequence of cell-cell diversity and the dynamic response of the cell population to treatment.

Resensitization due to redistribution of cells in the phases of the cell cycle during arbitrary radiation protocols.

When a cell population in exponential growth is subjected to ionizing radiation, the degree to which its long-term size is attenuated, relative to a control population that is not irradiated, depends not only on the total dose but also on the time pattern of dose delivery. Using a standard mathematical model for cycling cell populations with age-dependent radiosensitivity, it has recently been shown that normal progression of cells through the cycle tends to decrease this relative population size when the total dose delivery time is increased from essentially zero times to short, finite times (Chen et al., Math. Biosci. 126, 147-170, 1995). This mathematical result is an agreement with intuitive arguments and experiments long known in radiobiology. Mechanistically, it says that after the first part of a dose has preferentially eliminated the more sensitive cells of an exponentially cycling cell population, cell cycle progression, with the consequent redistribution of cells among cycle phases, tends to "resensitize" that population, an affect countering that of sublethal damage repair. The present paper now generalizes this result, demonstrating that the redistribution-induced increase of cell killing carries over to doses of arbitrary duration. That is to say, delivering a given dose over some extended period will result in lesser ultimate population size (i.e. population size measured at some fixed time long after irradiation has ceased) than will delivering the same total dose acutely. The redistribution-induced resensitization occurs no matter how radiosensitivity depends on cell age. For illustration, examples are given to show that, for a split dose, the least sensitivity is observed when the two doses coincide. These examples also demonstrate, within the constraints of the overall resensitization principle, the possibility of an oscillatory dependence of population sensitivity on interfraction time.

A Monte Carlo/Markov chain model for the association of data for chromosome aberrations and formation of micronuclei.

The micronucleus assay is a convenient, in situ method for observing cell damage resulting from exposure to clastogenic agents and has been widely used as a dosimeter of human exposure to radiation or chemicals. It also is a complement to the classic clonogenic cell survival assay in that it can be used to examine radiation damage vs dose as a function of cell type or radiation quality. Digitized imaging densitometry was conducted on CHO cells that have undergone one division and in which further cytokinesis was blocked to collect data on the distributions of percentage total cellular DNA per micronucleus and frequency of micronuclei per cell after gamma irradiation in G1 phase. Theoretical counterparts to both classes of distributions were generated by a Monte Carlo double-strand breakage (DSB) simulation to the CHO genome, followed by simulated repair of this initial damage using a Markov chain algorithm that assumes linear restitutions of single DSBs complete with quadratic, incomplete exchanges among pairs of DSBs. Micronuclei were presumed to consist of single acentric fragments (including fused acentric pairs). The empirical distributions, when compared to their fitted theoretical counterparts, suggest, inter alia, that: (1) a slight dependence of micronucleus size on dose exists, with a trend toward higher density in the 2-4% genome range, at the expense of the 0-2% range, with increasing dose; (2) the probability of exchange incompleteness is at least 20%; and (3) the dispersions of the micronucleus frequency distributions are progressively lower than their (essentially constant) counterparts with increasing dose. Suggested is a cooperative increase in the number of fragments per micronucleus with increasing dose. Beyond these specific results, however, it is clear that furthering the understanding of the connection between DNA aberrations and formation of micronuclei would further link these two large bodies of data.

Assessment of radiation response on a cell-by-cell basis using in situ densitometric imaging of micronuclei.

Computerized densitometry of Feulgen-stained CHO cells was used to collect data on micronucleus DNA content as a function of radiation dose. A cytochalasin-B block technique was used to assess the DNA content of each micronucleus relative to the DNA in its own binucleated cell. Distributions were compiled on the number, percentage of cellular DNA per micronucleus, and total DNA loss to micronuclei per individual micronucleated cell as a function of dose. Nonparametric distribution summaries were used to compare response distributions as a function of dose. Mean micronucleus DNA content did not vary significantly with dose. A dose-dependent increase was noted in the fraction of micronucleated cells, the average number of micronuclei per micronucleated cell, and the percentage of total DNA in micronuclei per micronucleated cell. The data on percentage of cell DNA per micronucleus appear sufficiently detailed to be useful for studying mechanisms of chromosome fragment segregation and fusion during micronucleus formation.

Differences in the X-ray sensitivity of cells in different regions of the sandwich, a diffusion-limited system for cell growth.

The sandwich system was recently developed as a tumor analog; like spheroids, sandwiches are diffusion-limited multicellular systems which exhibit a necrotic center and a viable cell border. Using sandwiches of the 9L cell line, we compared the X-ray sensitivity of cells in the inner half of the viable border, adjacent to the necrotic center, with that of cells in the outer half of the viable border, adjacent ot the medium. No cells were hypoxic at the time of irradiation. The cells in the inner half of the viable border exhibited an increased radioresistance over cells in the outer half. The effect was dose multiplying with a multiplying factor of 1.5. Besides the sandwich studies, the X-ray sensitivity of 9L plateau monolayer cultures (induced by starvation) was compared to exponentially growing monolayer cultures. The plateau cultures exhibited an increased radioresistance over the exponentially growing cultures. The effect was also dose multiplying with a multiplying factor of 1.5.