Mechanisms of Methotrexate Resistance in Osteosarcoma¹

OS³ is the most common primary malignant bone tumor, mainly occurring in children and adolescents. Five-year disease-free survival has increased to >60% with current protocols, including a combination of limb salvage surgery and neoadjuvant chemotherapy (1, 2). High-dose MTX with leucovorin rescue is a major component of current protocols for the treatment of OS (1, 2, 3, 4).

MTX is a potent inhibitor of DHFR, a key enzyme for intracellular folate metabolism, which functions to regenerate tetrahydrofolate from dihydrofolate (5). In experimental systems, resistance to MTX can occur through a variety of mechanisms, including impaired transport of drug into the cell via the RFC, an increase in DHFR due to gene amplification or increased transcription, and diminished intracellular retention secondary to decreased polyglutamylation (5). In patients with acute myelocytic leukemia, a disease in which MTX is ineffective, the basis of intrinsic resistance is primarily a result of impaired polyglutamylation leading to lack of drug retention (6). In patients with relapsed acute lymphocytic leukemia, impaired transport and DHFR amplification are the common mechanisms of acquired resistance (7, 8, 9, 10).

Although high-dose MTX is frequently used in the treatment of OS, conventional dose therapy is ineffective. Several retrospective studies have suggested that a threshold peak MTX level needs to be achieved to obtain a good histological response to chemotherapy (11, 12, 13, 14). This relationship suggests OS is intrinsically resistant to conventional doses of MTX that may be overcome by the use of very high doses. To date, little information is available concerning intrinsic or acquired mechanisms of MTX resistance in this disease (15).

Our hypothesis is that alterations in transport through decreased RFC expression and increased DHFR expression are mechanisms of intrinsic and acquired MTX resistance in OS. These measures relate to patient outcome through determining the tumor’s response or resistance to chemotherapy. We have initially focused on the measurement of RFC and DHFR mRNA expression in fresh tumor samples by semiquantitative RT-PCR. The deletion or amplification of these genes was measured by quantitative Southern blot. Results of these assays are correlated with the histological response of the tumor to preoperative chemotherapy to determine their potential role as prognostic indicators.

Tumor samples were obtained at a single institution after obtaining written informed consent according to a protocol approved by the Memorial Hospital Institutional Review Board. All samples were confirmed to have a pathological diagnosis of high grade OS. A total of 42 specimens of OS were obtained from 1990 to 1995. The samples were obtained at the time of initial biopsy (n = 20), excision of the primary lesion which is an en bloc resection following a period of neoadjuvant chemotherapy (n = 13), and at recurrence (n = 8). Of the recurrent samples, six were pulmonary metastases and two were local recurrences. Of the eight patients with recurrent samples, all had received planned chemotherapy before relapse. The majority of patients were treated on the Memorial Sloan-Kettering T12 protocol or the pediatric Intergroup Phase III clinical trial, both of which have been described elsewhere (2, 16). Both of these protocols include multiple courses of high-dose MTX therapy. High-dose MTX is administered at a dose of 12 g/m² over 4 h, with leucovorin rescue initiated at 24 h. Most patients received four courses of MTX in addition to other chemotherapy, prior to excision of the primary lesion, which occurred 7–12 weeks after initial biopsy. Histological response to preoperative chemotherapy was determined by a single pathologist using the Huvos grading system as described previously (17, 18). Briefly, grade I indicates no evidence of necrosis, grade II indicates areas of necrotic material with other areas of histologically viable tumor, grade III indicates only scattered foci of viable tumor seen, and grade IV indicates no viable tumor seen in extensive sampling. Numerous studies have confirmed the histological response to preoperative chemotherapy as a prognostic indicator in OS, with patients having a grade III or IV histological response to preoperative chemotherapy having a significantly better event-free survival than patients with a grade I or II histological response (1, 2, 4, 12, 18). Most samples at the time of excision with a grade III or IV histological response were excluded because a piece of viable tumor tissue could not be identified for analysis.

All tumor tissues were snap-frozen in liquid nitrogen immediately after resection. Total RNA was isolated from the frozen tumor tissue using acid guanidinium-isothiocyanate (Tel-Test, Friendswood, TX), followed by phenol-chloroform extraction and isopropanol precipitation. RNA was resuspended in diethylpyrocarbonate-water, quantitated spectrophotometrically, and stored at −70°C. Genomic DNA from these samples was isolated by standard methods and quantitated spectrophotometrically (19).

The PCR primers for RFC, DHFR, and actin were synthesized by Operon Technologies (Alameda, CA), according to sequences published previously (8, 20). The RFC primers used were as follows: RFC617, 5′-CCAAGCGCAGCCTCTTCTTCAACC; and RFC949, 5′-CCAGCAGCGTGGAGGCAGCATCTGCC, which produce an ∼300-bp PCR product. The DHFR primers used were as follows: DHFR130, 5′ GTAGAAGGTAAACAGAATCTG; and DHFR505, 3′AGAACACCTGGGTATTCTGG. One μg of total RNA was reverse transcribed in a volume of 20 μl with 100 units of Superscript II RT (Life Technologies, Inc., Gaithersburg, MD) and 20 units of RNase inhibitor (Boehringer Mannheim, Indianapolis, IN) at 42°C for 60 min with random primers. After reverse transcription, the enzymes and first-strand cDNA were denatured at 95°C for 5 min and then chilled on ice for 5 min. Each cDNA sample was serially diluted with water as follows: 1:10, 1:10², 1:10³, 1:10⁴, and 1:10⁵ and added to a reaction mix including AmpliTaq DNA polymerase (Perkin-Elmer Cetus, Norwalk, CT), deoxynucleotide triphosphate, buffer, [³²P]dCTP, and each set of primers in a final volume of 25 μl. Amplification proceeded for 35 cycles of denaturation at 94°C for 1 min, annealing for 1 min, and elongation at 72°C for 1 min, with a final extension at 72°C for 5 min. The annealing temperature was different for each set of primers (RFC at 55°C and DHFR and actin at 60°C). Actin primers were used as a control for the amount of cDNA. The PCR products were electrophoresed on a 6% polyacrylamide gel subsequently dried on a gel dryer. The dried gel was exposed to a film overnight. RFC and DHFR expression was calculated by determining the ratio of RFC and DHFR relative to actin mRNA for each sample. A negative control in which no cDNA was added was included with each sample.

Aliquots (∼10 μg) of DNA were digested with EcoRI at 37°C for 5 to 20 h, separated on 0.9% agarose gels, and transferred onto nylon membranes using standard methodologies (21). After UV cross-linking for 5 min, the blots were prehybridized for at least 1 h and subsequently hybridized with a full-length DHFR or RFC cDNA probe, which was digested from a plasmid (RFC plasmid obtained as a kind gift from Wayne Flintoff, University of Western Ontario, London, Ontario, Canada). The cDNA was radiolabeled with [³²P]dCTP to a high specific activity using the random primer technique. The filters were washed at high stringency and directly scanned using a phosphor imager (Bio-Rad, Hercules, CA). The D12S2 probe was used as a loading control.

A χ² test was used to evaluate the difference between two variables.

Adequate RNA and DNA were obtained from 42 OS tumor samples. All tissue was confirmed by routine histological staining to be comprised of at least 80% tumor cells. The clinical data are summarized in Table 1. As mentioned previously, the majority of excision samples with a grade III or IV histological response were excluded because of an inability to identify tissue comprised predominantly of viable tumor cells. To investigate the transport status of MTX in OS, we measured expression of the RFC by semiquantitative RT-PCR (Fig. 1). Thirteen of 20 (65%) samples were found to have decreased RFC expression at the time of initial biopsy (Table 2). Decreased RFC expression was defined as a level below which no human leukemia blast sample had normal MTX transport, as determined by a functional assay in a previous study (8). The human lymphoblast CCRF-CEM cell line was analyzed with the OS samples to serve as a reference. The samples showed a wide range of RFC expression, but the majority of samples with low expression, as defined above, had barely detectable RFC mRNA (Fig. 2). Including samples from the time of definitive surgery and relapse as well as diagnosis, 23 of 42 (55%) were found to have decreased mRNA levels of the RFC. Seventeen of 26 (65%) samples with a poor histological response (grade I-II) to chemotherapy had decreased RFC expression, whereas 5 of 14 (36%) samples with a good response (III-IV) had a decrease (P = 0.03; Table 3). Among the biopsy samples, the seven with a poor histological response to chemotherapy had decreased RFC expression. To investigate whether RFC gene deletions or rearrangements were the basis of the decreased expression, we analyzed the samples by Southern blot. No evidence for a RFC gene deletion or rearrangement was observed in any of the samples (Fig. 3).

Because increased DHFR expression can result in MTX resistance, we also studied the DHFR mRNA levels in the OS samples by semiquantitative RT-PCR ((Fig. 4). Two of 20 samples (10%) showed increased DHFR expression at initial biopsy (Table 2). High DHFR was defined as expression levels higher than the CCRF-CEM cell line, which is sensitive to MTX. Increased DHFR expression was observed in 62% of patients (five of eight) with metastatic or recurrent disease (Table 2). The frequency of increased DHFR expression was significantly higher in these samples as compared to the biopsy material (P = 0.014; Table 3). To determine whether the increased DHFR expression was a result of gene amplification, Southern blots were performed. No evidence of DHFR gene amplification was found in any of the samples (Fig. 5).

We divided our samples into four groups based on the level of RFC and DHFR expression to look at the correlation between the expression of these two genes and response to chemotherapy. We found that patients whose tumors had normal levels of RFC and DHFR mRNA had the best response to chemotherapy (Fig. 6; P = 0.01). Ninety % of cases in this group had no metastatic disease.

DHFR gene amplification and impaired transport of MTX are two common mechanisms of acquired resistance observed in leukemia clinical samples (8, 9, 10). Although high-dose MTX is a commonly used chemotherapeutic agent in OS, there are few published studies on mechanisms of resistance (15). In this study, decreased RFC expression was found in 65% of OS biopsy samples. The high frequency of decreased RFC expression suggests that impaired transport of MTX may be an important mechanism of intrinsic resistance in OS. Southern blot analysis did not show evidence of RFC gene deletions or rearrangements, which suggests that the decreased expression may be related to decreased transcription or point mutations in this gene.

In OS, the only definite prognostic factor identified at diagnosis is the presence or absence of detectable metastatic disease (22). Some studies have suggested that P-glycoprotein, the product of the multidrug resistance (MDR-1) gene, may be a prognostic factor at diagnosis (23, 24), whereas other genes that affect therapeutic response have not been investigated. In this study, decreased RFC expression was associated with a worse histological response to preoperative chemotherapy that includes high-dose MTX treatment (P = 0.03). Histological response to preoperative chemotherapy correlates closely with event-free survival in OS (17, 18). It may therefore be worthwhile to investigate prospectively whether RFC expression at diagnosis predicts outcome.

Decreased RFC expression may partly explain why conventional dose MTX is ineffective in the treatment of OS because high doses may be needed to allow transport through alternative means, such as passive diffusion. The correlation of RFC expression with histological response to preoperative chemotherapy suggests that transport may not be the only determinant of intrinsic MTX resistance. In this study, MTX polyglutamylation was not investigated in the OS samples. Viable cells are required for these assays and were not available. Measurements of mRNA expression of folylpolyglutamate synthetase and γ-glutamyl hydrolase, the two enzymes responsible for determining MTX polyglutamate chain length, have not consistently correlated with direct determinations of MTX polyglutamylation (25). It is possible that high-dose MTX is effective in OS because it produces prolonged drug exposure. We are presently investigating MTX polyglutamylation in OS tumor samples.

Although increased expression of DHFR was rare in the biopsy material, it was frequent in the recurrent pulmonary metastases as well as the excision samples. Of the six recurrent metastatic lesions, four had increased DHFR expression. It is possible that the increased DHFR expression represents acquired MTX resistance. This could occur either through an acquired alteration in the tumor cells or through selection of a previously resistant clone. Previous studies in relapsed acute lymphocytic leukemia have demonstrated that approximately one-third of samples had increased levels of DHFR mRNA and enzyme activity (9). In relapsed acute lymphocytic leukemia, increased DHFR expression is associated with low-level gene amplification in ∼25% of patients (9). This differs from our observation in OS, where none of the samples had evidence of DHFR gene amplification. Other laboratories have made the observation that DHFRamplification is rare in OS tumor samples as well (15).

An alternative hypothesis to explain the high levels of DHFR expression in the pulmonary metastases is that it may reflect a difference between primary and metastatic tumors. Some reports have suggested the pulmonary metastases in OS are less responsive to chemotherapy than the primary site (12). The pulmonary metastases of colorectal metastases are less responsive to 5-fluorouracil than other sites of metastases or the primary tumor, most likely secondary to increased thymidylate synthase levels (26, 27). Increased thymidylate synthase levels in the pulmonary metastases of colon cancer are associated with increased E2F expression (28). Because E2F family members are also known to regulate DHFR transcription, elevated levels of E2F in the pulmonary metastases of OS may explain the high levels of DHFR expression (28). We are presently investigating the relationship of E2F and DHFR expression in the OS samples in the laboratory.

Lack of the retinoblastoma protein may play a role in MTX resistance in OS (29). In the absence of retinoblastoma protein, E2F levels increase, which results in an increase in transcription of several genes involved in DNA replication, including DHFR (30, 31). The retinoblastoma gene is frequently altered in OS, with loss of heterozygosity occurring in 75% of tumor samples (32, 33, 34, 35). We are also presently investigating the relationship of retinoblastoma pathway alterations in MTX resistance in OS.

The high frequency of decreased RFC expression in the OS biopsy samples may suggest new therapeutic strategies for the treatment of this disease. The high frequency of MTX transport impairments in relapsed acute lymphocytic leukemia led to studies of newer antifolates, such as trimetrexate, which do not depend on the RFC for cell entry. Leukemic cells that are resistant to MTX on the basis of transport are collaterally sensitive to trimetrexate, possibly due to decreased uptake of the natural folates (36, 37). To further widen the therapeutic index, trimetrexate can be administered with simultaneous leucovorin. Leucovorin, which uses the RFC for cell entry, protects the host but will not enter the cells with impaired MTX transport (38). In a Phase I trial of trimetrexate with simultaneous leucovorin, responses were seen in patients with OS (39). The high frequency of decreased RFC expression in the biopsy OS samples suggests that exploring this strategy further in patients with OS may be warranted.