Chapter 59:  Radioimmunotherapy of Cancer

Beta Emitters

Most RIT uses beta decay. In beta decay, a neutron breaks down, changing to a proton and emitting a high-energy electron (beta particle) and raising the atomic number by one without changing the mass number.^153,154 Given the length of their path, beta emissions are appropriate for treating tumors larger than 0.5 cm. In addition, not every cell needs to be targeted with a radionuclide conjugate. Bombardment of adjacent tumor cells by multiple beta particles results in enhanced killing through cross-fire, partially compensating for a lack of homogeneity of antigen expression from cell to cell. In theory, one might choose among beta emitters based on the size of the tumor. Shorter-range beta emitters such as ¹³¹I and ⁶⁷Cu might be used to treat micrometastatic disease, where a greater fraction of their decay energy would be deposited within small tumor cell clusters. Conversely, more energetic, longer-range beta emitters such as ⁹⁰Y could destroy larger tumor deposits and elminate tumor cells that had escaped direct targeting due to lack of antigen expression or poor vascularity.¹⁵⁴

Beta emission from radioisotopes kills tumor cells but also kills normal cells. As blood circulates through the bone marrow, beta decay from circulating radionuclide conjugates irradiates bone marrow cells producing myelosuppression. Sites of specific binding of radionuclide conjugates can also impact on myelotoxicity. In trials of RIT for lymphoma patients, greater radiation doses were delivered to bone marrow involved with lymphoma than to bone marrow that was lymphoma free.

¹³¹I was the first isotope used in radiotherapy, but it is not optimal for RIT of larger tumor deposits. ¹³¹I produces low energy beta particles, emits unwanted beta radiation, and exhibits a short biological half-life because of the action of tissue dehalogenases.¹⁵⁵ Myelosuppression can follow ¹³¹I-antibody treatment because of the radiation dose that the bone marrow receives from circulating conjugates. ⁹⁰Y emits only beta particles of appropriate energy for therapy but still exerts myelosuppression. The extent of heterogeneity of dose deposition in tumor is highly dependent on the antibody characteristics and radionuclide properties and can enhance therapeutic efficacy through the selective dose delivery to the radiosensitive areas of tumor. Radionuclide characteristics can affect the heterogeneity of dose deposition within viable and necrotic areas of a tumor. When ¹³¹I- and ⁹⁰Y-labeled radioconjugates were compared directly, ¹³¹I generally delivered a higher dose throughout the tumor, even though the instantaneous dose-rate distribution for ⁹⁰Y was more uniform.¹⁵⁶

Alpha-Emitters

Alpha emissions have high energies of several MeV, exhibit very short path lengths (< 100 μm), and are associated with a high probability of producing cytocidal DNA double-strand breaks. An individual cancer cell can be killed by interaction with only a few and possibly with only a single alpha particle.^154,157 Moreover, the path length of alpha particles is short enough to avoid damaging nontargeted regions. Homogeneous antibody distribution within a tumor is, however, essential if a bystander effect is to be observed on antigen-negative cells. A lack of homogeneous targeting may be more significant for solid tumors, which are often poorly vascularized and have high interstitial pressure, due to poor lymphatic drainage. Consequently, alpha emitters may be most effective in RIT directed against blood-borne tumor cells, micrometastatic disease, and cancer cells near the surface of cavities. Bismuth-212 and ²¹³Pb are attractive alpha-emitting radionuclides that are now available for clinical use. The ²¹²Pb precursor with a longer half-life can also be used to generate ²¹²Bi in vivo.¹⁵⁸ Another promising alpha-emitter, with a longer half-life of 7.2 hours, is ²¹¹At.

Use of alpha emitters in clinical trials has awaited developments in isotope production, protein-labeling chemistry, dosimetry, and the development of appropriate preclinical models.¹⁵⁴ Recent work with preclinical models suggests that RIT with alpha emitters can be more effective biologically than comparable doses of external beam irradiation or RIT with beta emitters. Bismuth-213-labeled anti-prostate-specific membrane antigen antibody has been used to target spheroids of human prostate cancer cells in culture as a model for prevascularized metastatic disease. A three-log reduction in spheroid volume was attained with RIT and micro-dosimetry suggested that the relative biologic effectiveness of the alpha-emitting conjugate was 3.2-fold that of external beam irradiation in the same system.¹⁵⁹ In a separate study, murine monoclonal CO 17-1A Fab' fragments were labeled with the alpha emitter ²¹³Bi or with the beta emitter ⁹⁰Y and each was used to treat immunosuppressed mice bearing human colon cancer xenografts. At equitoxic doses, RIT with the alpha-emitter linked to a monoclonal antibody fragment cured 95% of animals, whereas RIT with a beta-emitter achieved a cure rate of only 20%.¹⁶⁰

Zevalin (⁹⁰Y-iritumomab Tiuxetan)

Arming anti-CD20 antibodies with radionuclides has resulted in significant antitumor responses in patients with NHL. In one clinical trial, patients were randomized to treatment with either rituximab or ⁹⁰Y-iritumomab tiuxetan, a murine anti-CD20 monoclonal antibody covalently bound to tiuxetan, that forms a stable chelate with the pure beta-emitting radionuclide ⁹⁰Y. In a randomized trial that compared treatment with ⁹⁰Y-iritumomab tuxetan to treatment with rituximab in 143 patients with relapsed lymphoma, ⁹⁰Y-iritumomab tuxetan produced an 80% response rate compared to 56% with rituximab (p = .002).¹⁷³ Treatment of 54 patients with rituximab-refractory disease with ⁹⁰Y-iritumomab tuxetan resulted in an overall response rate of 74%, with 15% achieving a complete remission.

Radioimmunoconjugates can cause severe toxicities. For example, in a Phase I/II dose-escalation study of ⁹⁰Y-iritumomab tiuxetan, grade 3 and 4 hematologic toxicities were common. A dose of ⁹⁰Y-iritumomab tiuxetan containing 50 mCi of ⁹⁰Y was myeloablative and the patients required autologous stem cell support. Lower doses of ⁹⁰Y-iritumomab tiuxetan were still efficacious but were less toxic.

¹³¹I-tositumomab (Bexxar)

¹³¹I-tositumomab reacts with CD20 and emits both beta and gamma rays. In a pivotal study of 60 patients with chemotherapy-refractory low-grade or transformed low-grade NHL, treatment with ¹³¹I-tositumomab was compared with the patient's last qualifying chemotherapy (LQC).¹⁷⁴ The overall response rate to a single course of ¹³¹I-tositumomab was 65% compared with 28% with the patient's LQC (p < .001). Complete responses were observed in 20% after RIT and in 3% after the LQC (p < .001), with median durations of response of 6.5 and 3.4 months (p < .001), respectively. For complete responders, however, the median duration of response after the LQC was 6.1 months, whereas after RIT the median duration of response was more than 47 months. These results are comparable to those from an earlier study at the University of Michigan in which 7 of 53 patients remained in a complete response after ¹³¹I-tositumomab for 3 to 5.7 years.¹⁷⁵ Toxicities associated with this therapy included HAMA and flulike symptoms, which researchers attributed to the antibody. ¹³¹I-tositumomab has completed Phase III clinical trials in NHL, but the product has not yet been approved by the FDA for marketing.

Lym-1

Other lymphoma-associated antigens can serve as targets for RIT. Lym-1 is an IgG-2A that targets the β-chain of HLA-DR expressed on malignant and a fraction of normal lymphocytes. Lym-1 labeled with ¹³¹I has induced responses in 52% of 21 heavily pretreated patients with NHL who entered a trial, with complete remissions in 33% for a mean duration of 14 months.¹⁷⁶ Lym-1 has also been radiolabeled with ⁶⁷Cu or with ⁹⁰Y and the effect of the three different isotopes compared in clinical trials.^177,178 The ⁶⁷Cu-labeled construct, and less consistently the ⁹⁰Y-labeled construct, exhibited greater uptake and longer retention in tumor, resulted in a higher radiation dose, and improved the therapeutic index relative to ¹³¹I-labeled constructs.^178–180

Colorectal cancer

Several different onco-fetal and differentiation antigens have been targeted with RIT in human colorectal cancer. Behr and colleagues focused on targeting carcinoembryonic antigen (CEA) with ¹³¹I-labeled humanized antibodies with some objective responses.¹⁸³ I was also used to target the A33 antigen, a transmembrane glycoprotein of the immunoglobulin superfamily, with no significant responses.¹⁸⁴ Later, ¹³¹ I was replaced with ¹²⁵ I, producing a significant improvement in tumor retention.¹²¹ CC49 is a murine antibody that reacts with an adenocarcinoma antigen, TAG-72. ¹³¹I-CC49 and the mouse-human chimeric version of the antibody were used in clinical trials to treat patients with colorectal cancer, but they did not produce any significant clinical responses.¹⁸⁶ Because ¹³¹I-labeled antibodies can become rapidly dehalogenated in tumor tissue, the same antibody was labeled with ⁹⁰Y and used in a Phase I trial. Tissue biopsies of tumor targets and organs at risk were obtained to validate dosimetry estimates. The study detected only low tumor-normal tissue ratios and raised doubts about the potential of ⁹⁰Y-conjugated CC49 antibody for RIT of solid tumours.¹⁸⁷ Knox and colleagues reported the results of a Phase II clinical trial of ⁹⁰Y-DOTA-biotin pretargeted by NR-LU-10 antibody/streptavidin in patients with metastatic colon cancer.¹⁸⁸ Clinical responses were overshadowed by significant bowel toxicity resulting from the targeting antigen expressed by normal bowel.

Breast Cancer

The sensitivity of breast cancer to radiation has led to the development of tumor-targeted RIT. Antigenic targets include MUC1, CEA,and L6.^189–191 BrE-3 antibody, a murine IgG₁ monoclonal, reacts with an epitope on the tandem repeat of the peptide core of MUC-1.¹⁹² A Phase I trial was performed to explore the use of ⁹⁰Y-BrE-3 murine antibody. Although responses were observed, high titers or HAMA developed, preventing further use of the antibody.¹⁷⁶ The humanized version of the antibody has been evaluated in a clinical trial, and 8 of 17 patients (47%) showed responses, despite having failed previous conventional therapies.¹⁹⁴ Another anti-MUC1 antibody, m170, was radiolabeled with ⁹⁰Y and administered to patients.¹⁹⁵ Dosimetric studies have now progressed to treatment with measurable tumor regression and documented partial responses. Looking toward future trials, Kramer and colleagues have demonstrated synergy in an animal model of human breast cancer treated with a combination of humanized ⁹⁰Y-labeled BrE-3 and cytotoxic drugs such as capecitabine.¹⁹³

When first described by Gold and Freeman, CEA was thought to be a specific colon cancer marker.¹⁸⁸ Subsequent studies, however, demonstrated the expression of CEA in normal tissues and in cancers from other sites, including breast carcinomas.¹⁹⁵ NP-4, a well-characterized murine anti-CEA antibody, was labeled with ¹³¹I. In a Phase I/II study therapeutic responses were observed with ¹³¹I-NP-4. A total of 57 patients were treated with CEA-expressing tumors, mostly in very advanced stages. Modest antitumor activity was seen in 12 of 35 assessable patients with 1 partial remission, 4 minor/mixed responses and 7 instances of stabilization in previously rapidly progressing disease.¹⁹⁶

The L6 cell surface antigen is highly expressed in breast cancer.¹⁸⁹ This 24-kDa surface protein is related to a number of cell surface proteins with similar predicted membrane topology that have been implicated in cell growth. A chimeric antibody ChL6 labeled with ¹³¹I produced anecdotal tumor responses in patients with metastatic breast cancer who had failed standard therapy. DOTA chelation was used to arm the radioimminoconjugate with ⁹⁰Y (⁹⁰Y-DOTA-peptide-ChL6). Excellent tumor targeting was demonstrated in human tumor xenografts growing in immunosuppressed mice, and effective targeting was associated with an enhanced therapeutic index.¹⁹⁷ Efficacy of this molecule in combination with taxol demonstrated a synergistic effect, with 50% long-term survival in mice bearing human breast cancer xenografts.¹⁹⁸ In recent studies, Burke and colleagues examined in detail combined modality RIT with ⁹⁰Y-DOTA-peptide-ChL6 in xenograft models. When RIT was combined with an inhibitor of α_vβ₃-integrin, with paclitaxel, with a monoclonal antibody to EGFR, or with a bcl-2 antisense oligonucleotide, some combinations increased the cure rate, some decreased the cure rate, and some enhanced toxicity.¹⁹⁹ Although this supports the potential of combined modality RIT, it does emphasize the complexity of such systems.

Ovarian Cancer

Two tumor-associated monoclonal antibodies, HMFG1 and HMFG2, directed against MUC-1 and labeled with ¹²³I have been used to detect primary and metastatic ovarian, breast, and gastrointestinal neoplasms by immunoscintigraphy.^200,201 HMFG1 has been conjugated to ⁹⁰Y and used in a clinical study to treat patients with advanced ovarian cancer following conventional therapy.^202,203 Results for patients with minimal residual disease were encouraging, with half the patients being free from disease and in complete remission several years after treatment, consistent with the possible activity of RIT against minimal residual on the surface of the peritoneal cavity. Among 21 patients who achieved complete remission following surgery, chemotherapy, and intraperitoneal RIT, 78% survived for longer than 10 years, and the median survival had not been reached with a maximum follow-up of 12 years.^204,205 In a multi-institution international randomized concurrently controlled Phase III clinical trial, ⁹⁰Y-HMFG1 is now being evaluated; data accrual has recently been completed but has not yet been analyzed.

The response of patients with ovarian cancer following RIT with intravenously administered ¹³¹I-labeled chimeric monoclonal antibody MOv18 directed against a tumor-associated folate receptor is also currently being evaluated.²⁰⁶ To date, results have shown that ¹³¹I-labeled c-MOv18 can be given safely to patients with normal bone marrow reserves and may have therapeutic potential, particularly in patients with minimal residual disease.

Prostate Cancer

TAG-72 was one of the first antigens to be targeted with radiolabeled antibodies, but no major responses were observed in prostate cancer patients.^207,208 Prostate-specific membrane antigen radioimmunotargeting has also been evaluated.²⁰⁹ Disseminated prostate cancer has been treated with radiation and chemotherapy using doses of cytotoxic drugs that would otherwise not be tolerated with external beam radiation. Recently, O'Donnell and colleagues published the combined effects of a radioimmunoconjugate, ⁹⁰Y-DOTA-peptide-ChL6, with taxanes against human tumor xenografts in immunosuppressed mice.²¹⁰ They observed a 67% cure rate with the combination, whereas no mice were cured with RIT alone or chemotherapy alone. Comparable doses of each agent are achievable in humans and are expected to provide therapeutic synergy without increased toxicity. Recent studies with unconjugated ⁸⁹Sr suggest that targeting of bone with free isotopes can significantly prolong the survival of patients with androgen-independent prostate cancer.²¹¹ Consequently, targeting the site of most frequent metastasis might also have a role in RIT.

Renal Cancer

Metastatic renal carcinoma has been treated with a ¹³¹I-labeled mouse monoclonal antibody (G250).¹⁴⁰ Thirty-three patients with measurable metastatic renal cell carcinoma were treated in a study by Divgi and colleagues. There were no major responses. On the basis of external imaging, ¹³¹I-labeled mouse monoclonal antibody G250 showed excellent localization to all tumors that were 2 cm or more. Of the patients studied, 17 of 33 patients had stable disease, with tumor shrinkage observed in 2. Antibody immunogenicity restricted therapy to a single infusion.

Brain Tumors

Because brain tumors in adults rarely metastasize outside the cranium, locoregional approaches to RIT have been evaluated in patients with glioblastoma multiforme (GBM), a malignancy with an extremely poor prognosis that generally kills through local invasion and regional metastasis. Anti-tenascin antibodies have been radiolabeled, initially with ¹³¹I and more recently with ⁹⁰Y, and administered directly into the tumor.²¹³ Responses have been observed in 40% to 47% of patients. Median survival in patients with GBM was prolonged to 25 months (¹³¹I group) or 31 months (⁹⁰Y group) in an Italian trial.²¹³ In a separate study, the efficacy and toxicity of infusing ¹³¹I-labeled murine anti-tenascin antibody 81C6 directly into the resection cavity has been assessed in 33 patients with previously untreated malignant glioma.²¹⁴ Median survival for all glioma patients was 86.7 weeks, and for GBM patients, was 79.4 weeks, exceeding that for historic controls treated with conventional radiotherapy and chemotherapy postoperatively. Reversible hematologic toxicity was observed in 27% of patients and symptomatic neurologic toxicity occurred in 15%, but only a single patient required reoperation for radionecrosis. These observations suggest that a randomized phase III study would be of value. Other antigens, such as the EGFR variant III, could serve as unique targets on GBM, provided that linkers can be developed that minimize the release of isotope by intracellular processing of the internalized receptor.²¹⁵

A three-step approach has been developed to improve localization of radioconjugates to brain tumor cells. Twenty-four patients with recurrent high-grade glioma underwent a second surgical debulking and implantation of a catheter into the resection cavity. Biotinylated anti-tenascin antibody was infused initially. After 24 hours, avidin was administered, followed 18 hours later by ⁹⁰Y-biotin.²¹⁶ In a subsequent Phase II study, 37 patients with high-grade glioma, including 20 patients with GBM, were treated in a similar manner.²¹⁷ Among 12 historical controls with GBM, median survival was 8 months compared to 33.5 months in the pretargeted RIT group. Again, a randomized concurrently controlled trial is warranted.

Hyperfractionation

Recently, the concept of fractionating or hyperfractionating RIT by administering multiple doses of the radiolabeled antibody has been suggested to overcome heterogeneity of monoclonal antibody distribution in the tumor and the consequent nonuniformity of tumor radiation doses.²¹⁸ Preclinical data suggest that toxicity can be reduced, efficacy increased, dosimetry improved, and distribution made more uniform.²¹⁹ Clinical data have shown that toxicity can be reduced and the maximum tolerated dose can be increased for many patients.²¹⁹ Availability of humanized antibodies may improve the feasibility of this approach for immunocompetent patients.

Targeting Tumor Vasculature and Altering Tumor Hemodynamics

Recent studies suggest that tumor vessels have distinctive antigenic determinants that can be targeted with monoclonal antibodies (see Chapter 9, “Biochemistry of Cancer”). Use of beta emitters with appropriate energies and path lengths could deliver radiation to poorly vascularized sites. Increased perfusion of tumors and increased uptake of radionuclide conjugates has been achieved with hyperthermia, although optimal conditions in the clinic have not yet been defined.^{154,220–222}

Nonintravenous therapy

Success with RIT introduced into the resection cavities of patients with GBM suggest that nonintravenous administration should be further exploited. Intrathecal administration of ¹³¹I-labeled monoclonal antibodies has produced clinical responses in patients with leptomeningeal involvement from breast cancer, melanoma, lymphoma, and pinealblastoma.^223,224 In a Phase I trial with 31 patients, ¹³¹I-anti-tenascin and ¹³¹I-labeled anti-chondroitin proteoglycan sulfate monoclonal antibodies were evaluated. Five patients remained free from disease progression for more than 409 days, and one patient with leptomeningeal involvment from melanoma had disease stabilization for 4 years.^225,226 Catheters inserted sterotactically into brain parencyma could also be used to deliver radioimmunoconjugates.

Results from the randomized concurrently controlled trial of intraperitoneal ⁹⁰Y-HMFG1 in ovarian cancer patients who have had a complete response to conventional therapy should soon be available. If positive, this will provide an important proof of concept, although recurrence of extraperitoneal disease may limit this approach.

Pretargeting

Localization of radioimmunoconjugates to cancer cells might be enhanced through pretargeting.²²⁷ This approach typically requires two or three separate components. In one scheme, the antibody is first targeted to the tumor followed by clearance of the residual circulating antibody that is facilitated by a clearing agent. A radioactive agent is then administered for selective capture at the tumor site. Pretargeting of radionuclides to tumors is particularly attractive in that it has the potential to greatly reduce the systemic toxicity of conventional RIT. The problem of pretargeting strategies is their inherent complexity and the immunogenicity of the components, which are generally not of human origin.

Schemes for pretargeted RIT have occasionally used bispecific antibodies with specificities for both tumor and radionuclide chelator, but more commonly, the very high-affinity interaction between biotin and avidin or streptavidin has been exploited.²²⁸ An infused antibody-streptavidin conjugate or fusion protein is first allowed to localize to a tumor target. A clearing agent is then used to remove the remaining circulating conjugate. Delivery of a radionuclide is achieved with the use of a a biotinylated chelator. The chelator-radionuclide complex is either captured by the antibody-streptavidin bound to tumor cells or cleared rapidly through the kidney due to its low molecular weight. This approach has been used to target lymphomas with an anti-CD20-streptavidin fusion protein (B9E9), followed by infusion of an agent that clears circulating fusion protein followed by the subsequent injection of a ⁹⁰Y-biotin conjugate. Preliminary results from a Phase I trial indicate that the therapy is well tolerated. The estimated dose of ⁹⁰Y delivered to tumors was comparable to that previously achieved with directly radiolabeled whole anti-CD20. The estimated doses to the whole body and bone marrow were, however, significantly lower, suggesting that the dose of ⁹⁰Y might be escalated in future trials. A similar strategy has been planned to treat gastrointestinal cancers using an ACC49-streptavidin fusion protein that reacts with TAG-72.

Significant advantages of pretargeted therapy over conventional RIT include the much greater ratios of radioactivity in tumor versus nontumor tissues, thereby lowering the whole-body exposure. The immunogenicity of streptavidin might, however, prevent the repeated treatment cycles that may be required for effective therapy. In addition, a rather large quantity of radionuclide must be administered to capture a small fraction of the radionuclide at the tumor target site.

Recombinant Antibody Constructs

Use of humanized antibodies has reduced the HAMA response to radionuclide conjugates as well as to unmodified antibodies. Single chain Fv (scFv) constructs have been developed that contain the variable regions of a single heavy- and single light-chain linked to retain immunoreactivity. Radiolabeled scFv have provided improved tumor-to-normal tissue ratios, but the absolute magnitude of the scFv uptake in tumor is lower that that seen with intact IgG. Consequently, the scFv fragments may prove more useful for diagnostic than for therapeutic applications.¹⁵⁴ (scFv)₂ dimers have shown improved accumulation in tumor tissue.²²⁹ Constructs that contain linker peptides can also undergo spontaneous noncovalent dimerization. The clinical evaluation of such constructs is currently being assessed.

Alpha-Particle RIT

As described above, alpha emitters might be particularly effective in RIT directed against blood-borne tumor cells, micrometastatic disease, and cancer cells near the surface of cavities. Eighteen patients with refractory AML have been treated at Memorial Sloan-Kettering Cancer Center with ²¹³Bi-humanized anti-CD33 antibody. No significant extramedullary toxicity was seen. The radionuclide conjugates localized to sites of leukemic involvement. Fourteen (93%) of 15 evaluable patients had a reduction in circulating blasts and 14 (78%) of 18 patients had reduction in the percentage of bone marrow blasts.¹⁵⁷ In preclinical studies with lymphoma cells, ²¹¹At has been conjugated to rituximab and has demonstrated a very high tumor-to-normal cell toxicity ratio in cell culture, supporting the initiation of clinical trials.²³⁰ Extension of previous studies with intracavitary therapy for GBM can now be achieved because methods have been developed for labeling antibodies with the alpha emitter ²¹¹At for clinical use.²¹³ A Phase I trial has been initiated at Duke University with chimeric anti-tenascin antibody labeled with ²¹⁴At administered into surgically created cavities formed during resection of recurrent GBM.¹⁵⁴ Less than 0.5% of injected dose escaped from the resection cavity into the blood pool. Cavity interface-to-normal brain dose ratios were about 150 times higher than reported previously in a similar trial of ¹³¹I-labeled anti-tenascin antibody.²³²