Synergistic cytotoxicity with 2'-deoxy-5-azacytidine and topotecan in vitro and in vivo

Synergy, when it can be convincingly established, is an effective strategy for the development of novel drug combinations. We have evaluated the interaction between 2'-deoxy-5-azacytidine (DAC) and 9-dimethylaminomethyl-10-hydroxycamptothecin (topotecan) based on our hypothesis that DAC, through DNA hypomethylation, might increase the transcription of topoisomerase I (topo I) leading to increased sensitivity to topotecan. Five human tumor cell lines, A375 melanoma, DX-3 melanoma, DMS4C non-small cell lung carcinoma, UP-1 unknown primary adenocarcinoma, SN12C renal carcinoma, and the murine CT-26 tumor cell line, were studied. Drug interactions were assessed using the multiple drug effect analysis of Chou and Talalay (Chors, T-C, and Talalay, P. Adv. Enzyme Regul., 22:27-54, 1984.). A synergistic interaction was documented in four human cell lines and the murine CT-26 line. An antagonistic interaction was observed with the SN12C cell line. The toxicology and efficacy of this combination were analyzed using CT-26 in BALB/c mice. Various treatment schedules were studied, including: single doses of each agent; single sequential combination treatments where DAC was administered followed by topotecan 24 h later; and multiple sequential treatments where DAC and topotecan were administered on days 1, 2, 8, and 9. Efficacy studies showed that the single sequential combination of DAC (50 mg/kg) and topotecan (10 mg/kg) resulted in tumor growth delay as compared to single doses of DAC (50 mg/kg) or topotecan (10 mg/kg). When the multiple sequential combination schedule was used, the antitumor effect was more pronounced. In that experiment 50% of the control animals had tumors of 20 mm by day 28. For animals receiving a single sequential treatment with DAC and topotecan, the median time until the mean tumor size reached 20 mm was 38 days, and for the group with multiple sequential combination treatments the time was 51 days. Studies of the mechanism of the interaction showed that the activity of topotecan versus each cell line correlated with the topo I activity in nuclear extracts However, there was no correlation between topo I levels and synergy and no reproducible increase in topo I activity following exposure to DAC. Thus, while the exact mechanism of the interaction remains unclear, DAC can be effectively combined with topotecan to enhance antitumor activity.

Identification of ras oncogene mutations in the stool of patients with curable colorectal tumors

Colorectal (CR) tumors are usually curable if detected before metastasis. Because genetic alterations are associated with the development of these tumors, mutant genes may be found in the stool of individuals with CR neoplasms. The stools of nine patients whose tumors contained mutations of K-ras were analyzed. In eight of the nine cases, the ras mutations were detectable in DNA purified from the stool. These patients included those with benign and malignant neoplasms from proximal and distal colonic epithelium. Thus, colorectal tumors can be detected by a noninvasive method based on the molecular pathogenesis of the disease.

Clonal origin of bladder cancer

BACKGROUND

Patients with cancer of the urinary bladder often present with metachronous tumors, appearing at different times and at different sites in the bladder. This observation has been attributed to a "field defect" in the bladder that allows the independent transformation of epithelial cells at a number of sites. We tested this hypothesis using molecular genetic techniques.

METHODS

We examined 13 tumors from cystectomy specimens from four women, using a method that analyzes the pattern of X-chromosome inactivation to determine whether the tumors were derived from the same precursor cell. In addition, we analyzed allelic loss on autosomes to determine whether different tumors had the same genetic alterations. The alterations evaluated included the loss of chromosome 9q sequences (commonly found in superficial bladder tumors) and the loss of 17p and 18q sequences (usually found only in advanced tumors).

RESULTS

For each patient studied, all the tumors had inactivation of the same X chromosome, whereas normal bladder mucosa cells had random patterns of inactivation. Moreover, each tumor that could be evaluated from a given patient had lost the same allele on chromosome 9q, suggesting that the loss of this allele preceded the spread of neoplastic cells elsewhere in the bladder. The losses of chromosome 17p and 18q alleles, which are late events in tumor progression, were not common to different tumors from the same patient.

CONCLUSIONS

A number of bladder tumors can arise from the uncontrolled spread of a single transformed cell. These tumors can then grow independently with variable subsequent genetic alterations.

Studies on the mechanism of the synergistic interaction between 2'-deoxy-5-azacytidine and cisplatin

2'-Deoxy-5-azacytidine (5-aza-CdR) and cisplatin interact to produce synergistic cytotoxicity against many human tumor cell lines. Preliminary experiments designed to explore the mechanism of this synergy suggested a poor correlation between synergy and the degree of genomic hypomethylation measured following exposure to 5-aza-CdR. Subsequent studies using plasmid DNA suggested that rather than DNA hypomethylation, incorporation of 5-aza-CdR into DNA mediated increased cisplatin binding to DNA and could therefore be essential to the synergistic interaction between these two agents. In this series of experiments, we evaluated the degree of synergy with cisplatin produced against two human melanoma cell lines by two additional antimetabolites that were chosen on the basis of their biochemical properties. In addition, we investigated the synergy between 5-aza-CdR and cisplatin in parental and 5-aza-CdR-resistant murine cell lines, which differed in their sensitivity to 5-aza-CdR and DNA methylation status but incorporated similar amounts of 5-aza-CdR into DNA when exposed to this antimetabolite. In the studies testing additional antimetabolites, cytosine arabinoside, which is incorporated into DNA but does not hypomethylate it, produced synergy with cisplatin that was similar or superior to that obtained using 5-aza-CdR. With 3-deaza-adenosine, which is not incorporated into DNA but produces DNA hypomethylation through inhibition of S-adenosylhomocysteine hydrolase, a primarily antagonistic interaction was observed in the two cell lines studied. In the 5-aza-CdR-sensitive and -resistant cell lines, a very similar synergistic interaction was documented for 5-aza-CdR and cisplatin despite the significant difference observed in DNA methylation levels. Taken as a whole, these data suggest that DNA hypomethylation was not critical to the synergistic cytotoxicity produced by 5-aza-CdR and cisplatin. This finding suggests additional strategies that could further modulate this interaction.

Cytogenetic heterogeneity of genetically marked and metastatically competent "dominant" tumor cell clones

Examination of tumors usually shows them to consist of phenotypically and clonally heterogeneous cell subpopulations. On the other hand, previous studies from our laboratory have provided compelling evidence for the rapid evolution, or overgrowth, of single "dominant" clones during the course of primary tumor growth. Thus in one such study, syngeneic CBA mice were injected with a mixture of 50-100 different genetically tagged clones of a mouse mammary carcinoma called SP1. The vast majority of these clones were non-metastatic. The different clones were tagged by random integrations of foreign DNA using calcium phosphate-mediated transfection of the plasmid pSV2neo, the resultant primary tumors were found to consist of a single dominant clone, called B5, which was also shown to be metastatically competent. The detection of single dominant clones such as B5 in primary tumors can be reconciled with the concept of tumor cell heterogeneity if it could be shown that the dominant clone was in fact heterogeneous for other genetic or phenotypic markers, i.e., was homogeneous only for the plasmid-based genetic marker used for its detection. To study this question, we examined the karyotypes of several sublines of B5, two derived from a primary advanced tumor and one from a spontaneous lung metastasis. We indeed found evidence to support the existence of marked cellular heterogeneity within and between the three sublines examined. Thus, while all three retained common cytogenetic markers, each also expressed unique markers. Moreover, karyotypic heterogeneity within a given subline was observed. Thus the concept of clonal dominance of primary tumors by metastatically competent cell subpopulations is not incompatible with the concept of the cellular heterogeneity of tumors. The implications of the results are discussed.

Continuous infusion 5-fluorouracil, etoposide and cis-diamminedichloroplatinum in patients with metastatic carcinoma of unknown primary origin

We conducted a phase II clinical trial of 5-fluorouracil (5 day continuous infusion), cis-diamminedichloroplatinum and etoposide in previously untreated patients with metastatic carcinoma of unknown primary origin. Of the thirty-six evaluable patients (21 adenocarcinoma, 14 undifferentiated carcinoma and 1 squamous cell carcinoma), eight patients responded to this treatment (4 CR, 4 PR). Responses were seen in both soft tissue and visceral disease. Toxicity was significant and included grade III/IV myelosuppression in over 90% of patients treated. Non-hematologic toxicity included nausea/vomiting and stomatitis. Although the remissions obtained in this study appear to be durable (median duration of complete remission greater than 24 months), the regimen does not appear to offer significant advantages over other less toxic and more easily administered cisplatin-based combinations.

Identification of p53 gene mutations in bladder cancers and urine samples

Although bladder cancers are very common, little is known about their molecular pathogenesis. In this study, invasive bladder cancers were evaluated for the presence of gene mutations in the p53 suppressor gene. Of 18 tumors evaluated, 11 (61 percent) were found to have genetic alterations of p53. The alterations included ten point mutations resulting in single amino acid substitutions, and one 24-base pair deletion. In all but one case, the mutations were associated with chromosome 17p allelic deletions, leaving the cells with only mutant forms of the p53 gene products. Through the use of the polymerase chain reaction and oligomer-specific hybridization, p53 mutations were identified in 1 to 7 percent of the cells within the urine sediment of each of three patients tested. The p53 mutations are the first genetic alterations demonstrated to occur in a high proportion of primary invasive bladder cancers. Detection of such mutations ex vivo has clinical implications for monitoring individuals whose tumor cells are shed extracorporeally.

Reduced tumorigenicity of murine tumor cells secreting gamma-interferon is due to nonspecific host responses and is unrelated to class I major histocompatibility complex expression

Spontaneous SP1 murine adenocarcinoma cells transfected with the murine gamma-interferon (IFN-gamma) gene expressed IFN-gamma (SP1/IFN-gamma) failed to grow in syngeneic hosts and grew in nude mice. The rejection of SP1/IFN-gamma cells was related to the amount of IFN-gamma produced and appeared to be mediated primarily by nonspecific cellular mechanisms, although some role for T-cells in the afferent arm of this response is possible. SP1 cells are H2-Kk negative but express class I antigens when producing IFN-gamma. However, class I major histocompatibility complex (MHC) expression, while likely necessary, was insufficient in itself to prevent tumor growth since secretion of greater than 64 units/ml IFN-gamma was needed to inhibit tumorigenicity while only 8 units/ml IFN-gamma could induce class I antigens. Similar results were obtained with the murine colon carcinoma CT-26, a tumor that constitutively expresses class I MHC antigens, further supporting the contention that class I MHC expression is not essential for the rejection response induced by IFN-gamma. The failure of SP1/IFN-gamma cells to protect against a challenge with parent SP1 cells argues that factors other than IFN-gamma production or class I MHC expression are needed to induce a protective response against weakly or nonimmunogenic tumor cells.

Prolonged continuous acyclovir treatment of normal adults with frequently recurring genital herpes simplex virus infection. The Acyclovir Study Group

In this 3-year study of suppressive acyclovir for recurrent genital herpes, patients with more than six recurrences per year were randomized initially to 400 mg of acyclovir or placebo orally two times per day, with recurrences treated with 200 mg of acyclovir five times per day for 5 days. In the second year of the study, all patients received acyclovir as a daily suppressive or intermittent acute therapy; in the third year, all received daily acyclovir. Among 525 patients completing 3 study years, 289 received 3 years of suppressive therapy and 236 received 1 year of acute therapy followed by 2 years of suppressive therapy. Of those who completed the third year, 61% were recurrence free that year; 25% of the suppressive therapy-only group were recurrence free for all 3 years. The annual recurrence rate dropped from more than 12 recurrences per year at baseline to 1.0 (suppressive therapy) and 1.4 (acute and suppressive therapy) recurrences during the third year. No significant toxic effects were detected. Daily suppressive acyclovir therapy was effective and well tolerated.

Immunogenicity of a non-class I MHC expressing murine tumor transfected with the influenza virus hemagglutinin or murine interleukin-2 genes

The transfection of murine SP1 tumor cells with the hemagglutinin (HA) gene of influenza virus results, after fluorescent-activated cell sorting (FACS), in the selection of high-HA-expressing cell lines called H4A and H4B. Both lines fail to grow in syngeneic animals at doses that result in 100% tumor take of non-transfected tumor cells. Both grow in immunosuppressed mice. SP1 and H4A or H4B cells express few class I major histocompatibility complex (MHC) antigens but do express class II IAk antigens. H4A or H4B cells engender a cytotoxic T lymphocyte (CTL) response but cannot protect against a challenge with SP1 cells. This CTL response is inhibited by anti-CD4 but not anti-CD8 antibodies. Using FACS, we were able to select a population (called H5AK5) with high class-I MHC antigen expression. Like H4A and H4B, H5AK5 cells fail to grow in syngeneic animals but do grow in immunosuppressed mice. However, unlike H4A or H4B, H5AK5 can induce protection against a challenge with 1 x 10(5) SP1 cells. These studies indicate that the immunogenicity of HA-transfected SP1 cells may correlate with the cell-surface expression of class II MHC antigens. However, HA-expressing SP1 cells seem able to induce a protective response against a parent SP1 cell challenge only if they also express class I MHC antigens. This view is supported by the observations that SP1 cells expressing murine interleukin-2 do not express class I MHC antigens, fail to grow in syngeneic animals, do grow in immunosuppressed mice but do not protect against a challenge with parental SP1 cells.

Synergistic cytotoxicity using 2'-deoxy-5-azacytidine and cisplatin or 4-hydroperoxycyclophosphamide with human tumor cells

The combined use of 2'-deoxy-5-azacytidine with cisplatin or 4-hydroperoxycyclophosphamide in vitro frequently resulted in synergistic cytotoxicity against a panel of six human cell lines. This enhanced cell killing occurred at drug concentrations that are clinically achievable. Synergy was also seen if the sequence of drug administration was altered, implying that a temporal overlap between the drugs was necessary, but that the actual biochemical lesions induced by each agent were probably unique, and interacted in an as yet undefined manner. Of further interest was the observation that at least one of the synergistic pairs was active against five of the six cell lines tested. 2'-Deoxy-5-azacytidine incorporation as assessed by the level of gross genomic DNA methylation did not appear to correlate with the synergistic cytotoxicity observed. Thus, we could not discern a clear relationship between the degree of DNA hypomethylation and the observed synergies, although DNA hypomethylation frequently occurred when synergy was demonstrated. The practical usefulness of these drug combinations has not yet been tested and awaits appropriate clinical trials both to assess the tumoricidal effects and possible increased toxicity.

Transformation injury and the unicellular phenotype of malignant cells

In 1984, Philippe Shubik wrote in an editorial in the Journal of the National Cancer Institute that: 'In terms of general biology, the multicellular organisms have evolved a complex series of defensive responses to extracellular injury involving various inflammatory reactions and their systemic concomitants. Needless to say, these reactions are by no means always successful from the standpoint of the host and indeed may well be the immediate cause of the obvious ill effects noted. The unicellular organisms, in contrast, react to injury by dividing and moving. In the search for the features that may link chemical, physical, and viral carcinogens apart from their ability to induce neoplasia, only one characteristic in common is obvious, namely, their ability to produce intracellular change or injury while leaving the cell viable. Perhaps the initial and fundamental characteristics of neoplasia is a reversion of the cell to unicellular behavior. Division and invasiveness are the characteristics of the neoplastic cell, and increased motility certainly seems to be the most likely mechanism for invasion'. Dr. Shubik presented these views as a basis for 'further discussion' regarding the nature of the neoplastic response. We hope that this presentation will augment Shubik's plea by reviewing his idea in the context of our current knowledge of tumor development. In addition, we will attempt to integrate the concept of the unicellular behavior of tumor cells with Foulds' and subsequently Nowell's insightful hypothesis concerning tumor progression.

Phase I clinical trial of cisplatin given i.v. with 5-fluorouracil and high-dose folinic acid

We report the results of a phase I study of intravenously administered cisplatin, 5-fluorouracil and high-dose folinic acid. This trial was designed to exploit potential biochemical interactions between these three agents. The maximum tolerated doses were cisplatin, 75 mg/m2, day 1; 5-fluorouracil, 375 mg/m2, days 1-5 and leucovorin 500 mg/m2, days 1-5. The dose-limiting toxic effect of this regimen was myelosuppression. Mild non-hematologic toxic effects were also observed and included nausea, vomiting, stomatitis, and diarrhea. Phase II trial of this regimen are underway, however randomized studies will eventually be necessary to establish whether cisplatin contributes clinically significant activity to this regimen.

The use of a genotypic marker to demonstrate clonal dominance during the growth and metastasis of a human breast carcinoma in nude mice

When a mixture of 11 clones of a human breast carcinoma (MDA-MB-435)--each clone transfected with pSV2neo and identified as having a unique insertion site of the gene--was injected into nude mice, the resulting tumors were found to contain only one clone (Neo 24). This clone, identified by the unique restriction fragments on Southern blot analyses, was also found in metastases recovered from the lungs and lymph nodes of the animals. The individual clones showed no differences in in vitro growth, while in vivo the Neo 24 cells produced the largest tumors. Thus, one explanation for the observed clonal dominance in this study could be the more rapid growth in vivo of the Neo 24 cells. This study illustrates how an introduced selectable gene marker can be used in lineage studies of human tumor cell populations.

Interleukin-2 production by tumor cells bypasses T helper function in the generation of an antitumor response

A poorly immunogenic murine colon cancer was used to investigate mechanisms of antitumor immunity. Injection of tumor cells engineered by gene transfection to secrete IL-2 stimulated an MHC class I-restricted cytolytic T lymphocyte (CTL) response against the parental tumor. The tumor cells secreting IL-2 produced an antitumor response in vivo, even in the absence of CD4+ T cells. Animals immunized with the engineered cells were protected against subsequent challenge with the parental tumor cell line. Similar findings were demonstrated for other tumor types. Thus, provision of a helper lymphokine in a paracrine fashion induced a tumor-specific immune response involving activation of endogenous CTLs and other immune effector cells. These findings demonstrate that the failure of an effective antitumor immune response may be primarily due to a helper arm deficiency of the immune system rather than a paucity of tumor-specific cytotoxic effector cells. Furthermore, they outline a novel strategy for augmenting tumor immunity.

Nutritional support in hospitals in the United Kingdom: national survey 1988

The increasing sophistication of enteral and total parenteral nutritional support techniques has resulted in improvements in the clinical practice of such support in recent years. This survey was designed to establish current clinical practice in the management of nutritionally-compromised hospital patients in 206 districts in the United Kingdom. However, despite recent developments in these techniques, the response revealed a wide variation in the practice of clinical nutritional support. The results also indicate that in each district there should be a group of people, with an interest in clinical nutrition, to monitor and advise on nutritional support. It is suggested that, a national multidisciplinary group should be formed, similar to the American and European Societies of Parenteral and Enteral Nutrition. The group would promote the appropriate use of, and research into, nutritional support specifically for the nutritionally compromised patient in the United Kingdom.

Biochemical analysis of the role of transmethylation in the methionine dependence of tumor cells

The effects on methionine metabolism of the substitution of homocysteine for methionine in vitro were investigated in normal and tumor cell lines differing in their ability to utilize homocysteine for growth. The major finding of this study was that methionine-independent (Met-Indep) cell lines had much lower basal transmethylation rates than methionine-dependent (Met-Dep) cell lines. This was particularly evident in the parent SP1 cell line and its Met-Indep revertant, SP1-R. SP1-R compensated for a lack of methionine by reducing both its transmethylation and growth rates. An analysis of other potential differences in methionine metabolism between Met-Dep and Met-Indep cell lines failed to demonstrate any consistent abnormalities in all but the absolutely Met-Dep MDAY-D2 cell line. Thus, protein, S-adenosylmethionine, and polyamine synthesis were the same in Met-Dep and Met-Indep cell lines. These results indicate that the major regulatory step in determining the Met-Dep phenotype is an inherent increase in the rate of transmethylation reactions. Cell lines with high basal transmethylation rates cannot compensate for a relative deficiency of methionine and either cease growing (MDAY-D2) or generate revertants (SP1-R) for which the basal rate of transmethylation is considerably reduced.

Immunological characteristics of immunogenic variants induced in the MCA-F murine fibrosarcoma using 1-methyl-3-nitro-1-nitrosoguanidine, 5-aza-2'-deoxycytidine, or ultraviolet radiation

The purpose of this study was to compare the frequency of generation and in vivo cross-reactivity of highly immunogenic (Imm+) clones induced in a single parental murine fibrosarcoma cell line MCA-F by 4 weekly treatments with either UV-B radiation, 1-methyl-3-nitro-1-nitrosoguanidine, or 5-aza-2'-deoxycytidine. These agents are believed to induce Imm+ variants by different mechanisms. The frequency of Imm+ variant generation was similar for the three different protocols, suggesting that the frequency of Imm+ generation was related more closely to the cell line than the inducing agent used. The strength of the immunogenic phenotype, however, was better correlated to the agent used, since 1-methyl-3-nitro-1-nitrosoguanidine yielded clones with the strongest immunogenicities. Three of four UV-B-induced Imm+ clones grew preferentially in chronically UV-irradiated syngeneic mice, a phenotype associated with UV-induced skin tumors. Cross-reactivity was tested with two Imm+ clones from each treatment group in a modified immunoprotection assay that selectively engendered antivariant, but not antiparental, immunity. Under these conditions each clone, except one, protected against itself. The clones displayed a complex pattern of cross-protection. Intervariant cross-protection was sensitive to the challenge dose, suggesting possible differences in the strengths of the cross-reacting immunities. Conversely, parental cross-protection was observed only with high immunizing multiplicities of Imm+ cells. The clones expressed the Imm+ phenotype in both C3H/HeN and C3H/HeJ mice, suggesting that expression of mammary tumor virus antigens did not account for the strong antitumor immune response. We also investigated whether the level of major histocompatibility complex class 1 or class 2 expression and immunogenic phenotype were correlated. Flow cytofluorography using haplotype-specific anti-Kk and anti-Dk monoclonal antibodies did not reveal a consistent difference in the constitutive or gamma-interferon-induced class 1 expression by Imm+ clones. However, we did observe a significant increase in the constitutive expression of IAk by most of the Imm+ variant clones. Together, these data demonstrate that in this system Imm+ variants engendered by a variety of mechanisms can express a range of cross-reactive tumor rejection neoantigens, independent of parental tumor antigens or major histocompatibility complex antigen expression.

Unknown primary tumors: establishment of cell lines, identification of chromosomal abnormalities, and implications for a second type of tumor progression

We describe the establishment of two human unknown primary tumor (UPT) cell lines and a comparison of their cytogenetic properties with two direct harvests from two tumor biopsy specimens. Histopathological analysis of all four tumor specimens revealed them to be undifferentiated adenocarcinomas of unknown origin. The chromosome numbers in these samples vary between 38 and 144. Consistent structural anomalies involving chromosome 1 were observed in all cases. These included a deletion of the short arm (4 of 4), duplication of 1q (3 of 4) and the presence of homogeneously staining region (2 of 4). Several additional chromosomal changes involving chromosomes 7, 8, and 9 were also observed, but these were less consistent and their importance is not yet clear. Abnormalities in chromosome 1 have generally been associated with advanced malignancy. The finding of consistent chromosome 1 changes in UPT supports our hypothesis that these are type 2 progressors in that benign or less malignant stages are not readily identified. In short, UPT are likely malignant soon after transformation occurs, and this is reflected in their rapid acquisition of the metastatic phenotype and abnormalities in chromosome 1. We are therefore characterizing a number of additional UPT cases to determine if chromosome 1 and other cytogenetic changes are consistently associated with this unique subgroup of tumors.

Systemic immunity against a murine colon tumor (CT-26) produced by immunization with syngeneic cells expressing a transfected viral gene product

We have previously shown that CT-26 tumor cells expressing the transfected hemagglutination antigen of influenza virus elicit an immune response in syngeneic hosts that is cross-protective against an s.c. challenge with non-transfected parental cells. Since CT-26 is a murine colorectal carcinoma we sought to determine whether the same immuno-protective effect could be achieved after a challenge with parent tumor cells injected into the usual sites of colon tumor growth and metastasis. We now show that the protection afforded by this immunization protocol is systemic and that animals can be protected against tumor growth in the cecum, lymphatics of the mesentery, liver and lung.

Genotypic and phenotypic evidence of clonal interactions in murine tumor cells

The stability of mixed tumor cell populations has been described in terms of phenotypic characteristics such as metastatic potential, immunogenicity, and drug resistance. We have extended these analyses to the molecular level in a model that uses transfection of the hemagglutinin antigen (HA) gene of influenza virus into murine CT-26 colorectal carcinoma cells. Transfection was followed by fluorescence-activated cell sorting (FACS) to select a parent population with expression of high levels of HA. We characterized this population (FACS-3) and four derived clones (5, 6, 9, and 18) over time with regard to phenotypic characteristics: immunogenicity and cross-protection against tumor challenge, cell surface expression of HA, evidence of HA gene amplification, and levels of HA mRNA. During 6 months in culture, the FACS-3 parent cells remained stable, but the individual clones varied for all of the parameters assessed. Among the clones, all possible molecular variations occurred, including changes in HA gene copy number (increased in clone 5 and decreased in clone 18), gene rearrangement (clone 5), decrease in HA mRNA (clones 6, 9, and 18), increase in HA mRNA (clone 5), and an abnormality in translational control or a posttranslational error. In all cases, the molecular changes correlated with cell surface HA expression, immunogenicity, and cross-protective potential. We conclude that in vitro clonal interactions play a role in stabilizing heterogeneity in this system. These studies show that even in the absence of selection, clonal interactions may alter the phenotype of tumors by increasing malignant progression or impeding tumor growth.