HLA-haploidentical bone marrow transplantation with posttransplant cyclophosphamide expands the donor pool for patients with sickle cell disease

Sickle cell disease (SCD) is caused by a single nucleotide mutation in the sixth codon of the β-globin chain. The mutation leads to abnormal polymerization of hemoglobin (Hb) in response to deoxygenation, producing the characteristic sickle-shaped erythrocytes. SCD occurs in roughly 1 in 400 blacks in the United States.¹  The natural history of SCD is highly variable; however, most patients have a shortened life expectancy and experience significant morbidity that may include hemolytic anemia, severe pain, vaso-occlusive crises, stroke, avascular necrosis, pulmonary hypertension, infections, renal failure, and thrombosis. These complications often accrue despite the use of hydroxyurea and the judicious use of blood transfusions.^2,3 

Myeloablative allogeneic blood or marrow transplantation (BMT) from a histocompatible (HLA-matched) donor can cure SCD.^4,5  Adult patients with SCD have generally been excluded from myeloablative BMT trials because of anticipated excess morbidity and mortality resulting from accumulated disease-related end-organ damage. Efforts to use nonmyeloablative BMT strategies in adults with SCD were initially disappointing.^6,7  However, Hsieh et al reported on a nonmyeloablative schema that resulted in sustained donor engraftment in 9/10 patients, with no cases of acute or chronic GVHD. However, only 24 of 112 patients had matched related donors.⁸  Of these patients, major ABO incompatibility was felt to preclude BMT in 4, and only 10 patients (8.9%) actually proceeded to BMT.⁸  It is equally difficult to find HLA-matched donors for blacks in unrelated registries.⁹  Accordingly, broader application of the BMT in severe hemoglobinopathies is dependent on novel strategies that address the issue of limited donor availability.^10-12 

Our group and others have shown that high-dose cyclophosphamide early post-BMT effectively modulates alloreactivity associated with partially matched donors in animals and humans.^13-15  High-dose cyclophosphamide is highly toxic to lymphocytes but spares hematopoietic stem cells because of their high levels of aldehyde dehydrogenase, an enzyme responsible for metabolizing the drug.¹⁶  The strategy of giving high-dose cyclophosphamide on days 3 and 4 post-BMT also takes advantage of the heightened cytotoxic sensitivity of proliferating, alloreactive T cells over nonalloreactive, resting T cells.¹⁷  As a result, related HLA-haploidentical BMT using high-dose posttransplant cyclophosphamide has displayed similar levels of GVHD and long-term immune reconstitution as seen with matched sibling donor BMT.¹⁴  Our data in patients with hematologic malignancies demonstrating that haploidentical related donor BMT with high-dose posttransplant cyclophosphamide is safe and effective were confirmed in a recent cooperative group trial (BMT CTN 0603).^14,18 

Reduction of GVHD, and potentially a graft-versus-tumor effect, is often associated with an increase in relapse in patients with hematologic malignancies; however, a graft-versus-tumor effect is not necessary in nonmalignant diseases. Here, we tested the hypothesis that nonmyeloablative conditioning with high-dose posttransplant cyclophosphamide would expand the number of SCD patients eligible for allogeneic BMT, by allowing the safe and effective use of related haploidentical donors.

Patients aged 2 to 70 years and receiving their first BMT were eligible. Additional eligibility criteria included good performance status (Eastern Cooperative Oncology Group 0 or 1; Karnofsky and Lansky 70-100), ability to sign consent (or assent if minors) forms, and the presence of a fully or partially (at least haploidentical) HLA-matched first degree relative willing to donate. Eligible diagnoses included the following: sickle cell anemia (Hb SS), Hb S/β° thalassemia, Hb S/ β+ thalassemia, Hb SC disease, Hb SE disease, Hb SD disease, Hb SO-Arab disease, or Hb S/hereditary persistence of fetal Hb. In addition, patients had to have at least 1 of the following hemoglobinopathy-related complications already published as indications for BMT in these patients^4,8,19 : stroke, magnetic resonance imaging changes indicative of brain parenchymal damage, magnetic resonance angiogram evidence of cerebrovascular disease, acute chest syndrome requiring exchange transfusion or hospitalization, recurrent vaso-occlusive pain crisis (> 2 per year for the last 2 years), stage I or II sickle lung disease,²⁰  sickle retinopathy, osteonecrosis, red cell alloimmunization (> 2 antibodies) during long-term transfusion, constellation of dactylitis in the first year of life and a baseline hematocrit of 21% and leukocytosis (> 13.4 × 10³ mm³) in the absence of infection during the second year of life, history of invasive pneumococcal disease, pitted red blood cell count > 3.5% during the first year of life, abnormal transcranial Doppler, or transfusion dependence. The protocol (Johns Hopkins trial J0676) was approved by the Johns Hopkins Institutional Review Board (IRB), and patients signed an IRB-approved informed consent form before BMT. The study was registered at ClinicalTrials.gov number NCT00489281. In total, 14 patients received treatment on a phase 1/2 clinical trial after informed consent was granted in accordance with the Declaration of Helsinki. An additional 3 patients consented to the transplant, but they received identical treatment outside of the study because either their health insurance did not include a clinical trial benefit or they had an HLA-matched donor at the time when this study was only open to haploidentical donors (see “Protocol evolution”). Permission to include these patients in the analysis of outcomes was granted by the Johns Hopkins IRB.

First degree relatives who shared at least 1 HLA haplotype with the patient, did not have SCD or another hemoglobinopathy, and were in good health were allowed to serve as donors. Patients with sickle cell trait were not excluded as donors. When more than 1 donor was available, the donor with the fewest HLA allele mismatches was chosen, unless the patient had donor anti-HLA antibodies or there was a medical reason to exclude the donor. If donor anti-HLA antibodies were detected the next best related match was chosen. Donor bone marrow was harvested with a target yield of 4 × 10⁸ nucleated cells/kg recipient ideal body weight and infused on day 0. The marrow was unmanipulated except that major incompatible ABO grafts had red blood cells depleted by buffy coat preparation and minor ABO incompatible grafts had plasma removed.

HLA phenotyping was performed as described previously.¹⁴  Potential family members were initially typed at the HLA-A, HLA-B, and HLA-DRB1 loci at an intermediate resolution level. Family members selected as donors were then further typed at the HLA-C locus at an intermediate resolution level. DRB1 and DQB1 alleles were typed at a high-resolution level. As needed, recipients and potential donors were typed at a high-resolution level for HLA-Cw alleles. Haplotypes were determined based on family studies whenever possible.

Patients received antithymocyte globulin ([ATG]; rabbit) 0.5 mg/kg on day −9 and 2 mg/kg on days −8 and −7; fludarabine 30 mg/m² on days −6 to −2, cyclophosphamide 14.5 mg/kg on days −6 and −5, and total body irradiation 2 Gy on day −1. Unmanipulated bone marrow was collected and infused on day 0. GVHD prophylaxis consisted of cyclophosphamide 50 mg/kg/d on days +3 and +4, and tacrolimus to maintain a level of 5 to 15 ng/dL (for 1 year) and mycophenolate mofetil 1 g every 8 hours (until day 35) were started on day 5 (Figure 1).

The first 2 patients were transplanted without ATG that was added by a protocol amendment after the report by Bernaudin et al suggested it may improve engraftment.⁴  Sirolimus replaced tacrolimus by protocol amendment starting on patient 11 in hopes of reducing the incidence of posterior reversible encephalopathy syndrome (PRES). The study was initially only open to patients with HLA-haploidentical donors but was amended to also include HLA-matched donors. Patients 15 to 17 received G-CSF–primed bone marrow grafts in and attempt to improve engraftment (G-CSF at 10 μg/kg/day subcutaneously for 5 consecutive days until the day before the harvest [ie, days −5, −4, −3, −2, and −1]). On these donors, the nucleated cell target range was targeted between 8 to 16 × 10⁸/kg recipient ideal body weight, with the volume not to exceed 20 mL/kg donor's weight once the minimal target of 8 × 10⁸/kg was reached.

All patients were started on penicillin V-K, with the recommendation to continue it lifelong. Standard broad-spectrum bacterial antibiotic prophylaxis with a quinolone and antifungal prophylaxis was given to patients who became neutropenic. All patients were on levetiracetam for as long as they were on immunosuppression. Patients received standard Pneumocystis jiroveci and anti-HSV/VZ prophylaxis for 1 year. Magnesium levels were kept above 1.5 mg/dL. All blood products, except for the allograft, were irradiated with 25 Gy before transfusion. The thresholds of RBC and platelet transfusions were hematocrit < 25% and platelet count < 50 × 10³/mm³. Cytomegalovirus (CMV)–seronegative patients were given transfusions from CMV-seronegative donors, or leukocyte reduced blood products if CMV-negative products were unavailable. Patients were monitored for CMV reactivation by weekly measurement of CMV copy number by PCR of serum until day 60. Preemptive therapy with ganciclovir was initiated when ≥ 300 copies of CMV/mL serum were detected. All patients' crises were evaluated by the pain service before proceeding to develop a plan in case of a crisis during treatment.

Patients had chimerism studies done on peripheral blood on days 30, 60, 180, and 360, and yearly thereafter. Chimerism was measured by PCR analysis of variable number of nucleotide tandem repeats unique to donors or recipients on total peripheral blood and isolated CD3⁺ T cells. Graft failure was defined as undetectable DNA of donor origin on at least 2 occasions no less than 1 week apart.

Between 2006 and 2011, 19 (17 adult and 2 pediatric) patients with SCD were referred for BMT and HLA typed. All patients had at least 1 unaffected HLA-haploidentical donor. Two patients were not offered BMT: one because of high levels of donor-specific anti-HLA alloantibodies against his only related donor, and the other because the 2 potential donors were not medically eligible to undergo bone marrow harvest. Therefore, 17 patients (89%) proceeded to BMT (Table 1), 3 with suitable HLA-matched sibling donors and 14 with a HLA-haploidentical related donors (Table 2). The median age of the patients was 30 (range, 15-46 years). The 17 patients had SCD (14 with Hb SS, 2 with Hb SC, and 1 with Hb SS and α-thalassemia) and were suffering frequent and severe pain crises. All patients (except patient 4) were treated with hydroxyurea with suboptimal results. Although not required by protocol, exchange transfusions were used before the BMT to reduce HbS levels to < 30% in 12 patients (patients 2, 5-12, 14, 15, and 17) as published previously.⁸  Patient 1 data were reported previously.¹³  Data presented are as of June 30, 2012.

In the 14 patients not receiving G-CSF–primed marrow, the marrow grafts had a median CD3⁺ cell count of 3.61 × 10⁷/kg recipient ideal body weight (range, 3.29-6.59 × 10⁷), a median CD34⁺ cell count of 5.24 × 10⁶/kg recipient ideal body weight (range, 4.4-8.68 × 10⁶), and a median nucleated cell count of 4.35 × 10⁸/kg recipient ideal body weight (range, 4.6-5.2 × 10⁸). In the 3 patients receiving G-CSF–primed bone marrow, a median CD3⁺ cell count of 1.09 × 10⁸/kg recipient ideal body weight (range, 4.06 × 10⁷-1.63 × 10⁸), a median CD34⁺ cell count of 6.02 × 10⁶/kg recipient ideal body weight (range, 4.1-8.85 × 10⁶), and a median nucleated cell count of 1.14 × 10⁹/kg recipient ideal body weight (range, 1.04-2.16 × 10⁹). All patients recovered blood counts. The median time to neutrophil recovery over 500 × 10³/mm³ was 24 days (range, 0-35 days) and to last platelet transfusion to keep platelets counts over 50 × 10³/mm³ was 24 days (range, 0-61 days). Graft failure was not seen in HLA-matched patients, but 43% of the haploidentical patients rejected their graft. Patient 8 experienced primary graft failure, never demonstrating detectable donor DNA. Secondary graft failure was seen in 5 patients (2, 5, 12, 14, and 15), who initially showed evidence of donor engraftment but eventually completely reconstituted with host hematopoiesis at a median time to failure of 151 days (range, 62-1004 days). Table 3 shows the chimerism values obtained in peripheral blood. Ten engrafted patients converted to the Hb types and levels of their donors (Table 4). Patient 5 had late graft failure (day +1004) but before that had persistently low levels of donor DNA and was fully symptomatic with sickle cell–related problems (pain crises and transfusion dependence). At last follow-up, 6 patients, all recipients of haploidentical related allografts, have full myeloid chimerism and are off immunosuppression (Table 3). Patient 17 (a G-CSF–mobilized marrow recipient) had his immunosupression stopped approximately day 220 (see “Complications”) while fully engrafted (Table 3). The reasons patients are still on immunosuppression is either they are < 1 year from BMT (per protocol, patient 16) or because they are mixed chimeras (the rest).

All 17 patients are alive at a median follow-up of 711 days (range, 224-1981 days). Eleven patients showed evidence of long-term engraftment at levels sufficient to reverse their sickle cell phenotype and eliminate pain crises. One of these eleven patients (patient 9) has low levels of donor chimerism, has been transfusion independent, and has not been admitted to the hospital or experienced any SCD crises since the BMT > 700 days ago. Accordingly, the markers of anemia and hemolysis also improved in the group of engrafting patients (Table 4). In the 11 engrafting patients, the median bilirubin decreased from 2.3 to 0.4 mg/dL (reference 0.1-1.2 mg/dL, in haploidentical pairs bilirubin changed from 1.9 mg/dL to 0.3 mg/dL and in HLA-matched pairs it changed from 4.9 mg/dL to 0.6 mg/dL); the median reticulocyte counts decreased from 229.4 to 73.7 × 10³/mm³ (reference 24.1-87.7 × 10³/mm³, in haploidentical pairs reticulocyte counts changed from 232 × 10³/mm³ to 79.2 × 10³/mm³ and in HLA-matched pairs it changed from 210.4 × 10³/mm³ to 55.7 × 10³/mm³); and the median Hb increased from 10.1 g/dL (patients pre-BMT were transfusion supported) to 11.50 g/dL (reference 12-16.3 g/dL; in haploidentical pairs Hb levels changed from 9.8 g/dL to 11.4 g/dL and in HLA-matched pairs it changed from 10.1 g/dL to 11.5 g/dL). Before BMT, all patients were receiving regular transfusions as part of their therapy. Among the 11 engrafting patients, 10 became transfusion independent before day 60; patient 13 required red cells transfusions between days +60 and +180, but she has been transfusion independent since day 174 post-BMT.

All patients had pain early post-BMT. This was despite high levels of peripheral blood chimerism, improved Hb levels, and Hb S levels similar to their donors. Three engrafting patients had hospital admissions at our center after day 90 for pain control: patient 6 (last admission 12 months post-BMT), patient 7 (last admission 3 months post-BMT), and patient 10 (65 days post-BMT). The frequency and intensity of the painful crises improved after 3 to 6 months, and pain medicines were eventually tapered in 10 of the 11 patients with evidence of engraftment. None of the remaining patients with evidence of engraftment have acute vaso-oclusive pain. As of last visit, patients 1, 4, 9, 10, 13, 16, and 17 no longer require narcotic medications for pain. No clinical evidence of new cerebrovascular events, acute chest syndrome, or priapism has been recorded on engrafting patients.

None of the patients receiving haploidentical grafts developed GVHD. One of the patients who received a matched sibling transplant (patient 11), developed acute GVHD that involved only the skin on the face (< 9% body surface area). His GVHD resolved without systemic therapy; it was not biopsied because it only involved his face. No patients have developed chronic GVHD. Patients 2, 3, and 10 developed PRES, and all patients recovered fully. Patient 10 was switched from tacrolimus to sirolimus, and patients 11 to 17 received sirolimus by a protocol amendment, and none have developed PRES. Patient 6 developed a dental abscess on day −3, so the remaining conditioning and BMT were aborted; she was reconditioned (without repeat ATG) and transplanted 3 weeks later. Patient 13 had a seizure associated with noncompliance with levetiracetam. The drug was restarted, and no new seizure activity has occurred. Patients 9, 14, and 15 had CMV reactivation but not CMV disease. Patient 9 had an episode of Epstein-Barr virus viremia that responded to a single dose of rituximab. Patient 14 developed a respiratory syncytial virus upper respiratory tract infection and was found to have Mycobacterium tuberculosis on bronchoscopy; she recovered fully. Patient 17 was admitted 6 months post-BMT for the treatment pulmonary infiltrates of unknown etiology. Sirolimus toxicity was suspected, but fungal infection could not be ruled out.

BMT remains the only curative therapy for patients with SCD; however, most protocols have used myeloablative conditioning, excluded adult patients, and required HLA-matched sibling donors.^4,5  Because most patients with SCD will not have an unaffected matched donor, we sought to develop a safe and effective nonmyeloablative-related haploidentical BMT platform using posttransplant cyclophosphamide for both adult and pediatric patients. All 19 SCD patients referred to us for typing had at least 1 unaffected HLA-haploidentical related donor, and 17 (89%) were able to proceed to BMT. Long-term hematopoietic engraftment and abatement of sickle cell pain exacerbations was achieved in 11 patients (58% of those referred for BMT). Hsieh et al reported success in transplanting adult SCD patients using nonmyeloablative conditioning and HLA-matched sibling donors.⁸  In that study, 24 of 112 patients (21%) were eligible to proceed to matched sibling BMT; 10 patients (8.9%) were transplanted, and 9 (8% of those referred for BMT) experienced sustained engraftment and resolution of their sickle cell pain crises.⁸  Thus, by expanding the donor pool to include HLA-haploidentical donors, we have increased the percentage of SCD patients who may benefit from allogeneic BMT.

Historically, mismatched allogeneic BMT has been associated with mortality rates in excess of 50%, mostly as a result of GVHD.^21,22  We developed our haploidentical donor platform using bone marrow, rather than peripheral blood, in an attempt to limit GVHD for a procedure that had been previously associated with unacceptably high rates of GVHD.¹⁴  Using this platform, we have seen a rate of GVHD in haploidentical related BMT for malignancies similar to that seen with matched sibling transplants.^14,18  GVHD prophylaxis with high-dose cyclophosphamide has proven effective with very low rates of both acute and chronic GVHD after HLA-haploidentical related BMT, not only in single center studies but also in a recent multicenter study.^14,18  Only 1 case of self-limited acute GVHD involving just the skin of the face was seen in our 17 patients; in fact, no GVHD was seen in our haploidentical recipients, demonstrating that our GVHD prophylaxis regimen was highly effective. It is likely that the addition of ATG for the purpose of improving engraftment also helped to decrease the risk of GVHD.

Nonmyeloablative conditioning seems to be safe in adult patients with severe hemoglobinopathies who have accumulated significant end-organ damage.^23,24  However, nonmyeloablative regimens, including ours, have been associated with a high frequency of graft failure in patients with severe hemoglobinopathies, even when using matched sibling donors.^6,7  None of the patients who ultimately engrafted in the report by Hsieh et al⁸  achieved full donor chimerism, and all were still on immunosuppression at the time of publication in an attempt to prevent late graft failure. There are likely several reasons for the incomplete engraftment after nonmyeloablative conditioning for SCD, including the lack of cytotoxic therapy before BMT as well as the high number of transfusions that may induce alloimmunization.²⁵  Moreover, engraftment and GVHD are intimately linked.²⁶  Approaches that limit GVHD are usually associated with higher rates of mixed chimerism and graft failure.²⁶  Not only was limiting transplant-related mortality and severe morbidity our highest priority in these patients who can live decades with their underlying hemoglobinopathy but also antitumor activity associated with GVHD was not required. Thus, although graft failure is also undesirable, we believe it is preferable to GVHD-related mortality. Indeed, all patients who lost their graft recovered autologous hematopoiesis. Although the numbers are small, the 2 patients with ABO incompatible haploidentical transplants either rejected the graft (patient 14) or have low donor chimerism (patient 9).

In our study, 5 patients are full donor chimeras off all immunosuppression post-BMT (patients 1, 3, 7, 10, and 17, all of them haploidentical), whereas 1 patient is > 95% donor at 6 months (patient 16, haploidentical, too), 5 remain mixed chimeras (patients 4, 6, 9, 11, and 13, 3 are HLA-matched and 2 are haploidentical), and 6 have lost donor engraftment (patients 2, 5, 8, 12, 14, and 15). We continue to study approaches for improving engraftment without increasing GVHD. The higher cell counts associated with mobilized peripheral blood grafts seem to improve engraftment after nonmyeloablative conditioning.²⁷  Mobilized peripheral blood haploidentical allografts have been successfully used with posttransplant cyclophosphamide but seem to be associated with a higher rate of GVHD than bone marrow.²⁸  The last 3 patients on protocol received G-CSF–treated bone marrow; such bone marrow may improve engraftment without increasing T-cell numbers.²⁹  However, there are too few patients to conclude whether the administration of G-CSF was beneficial, and 1 of the 3 patients experienced a late graft failure. We have recently shown that, similarly to renal transplantation,³⁰  desensitization can prevent graft failure after haploidentical related donors, resulting from donor-specific anti-HLA antibodies.³¹  Although none of these patients had detectable anti-HLA antibodies against their donors, it is possible that alloimmunization related to the frequent transfusions was still present. Thus, desensitization even in patients without donor specific anti-HLA antibodies could prove useful.

The conditioning regimen was well tolerated, but some significant toxicities were encountered. PRES, a well-recognized complication of BMT in SCD,⁴  occurred in 3 patients, and resolved without neurologic sequelae. PRES was not encountered in the patients who received sirolimus instead of tacrolimus. All the patients developed at least 1 pain event at some point during the BMT course, stressing the need for a multidisciplinary approach in the care of these patients. Prolonged need for narcotic use posttransplant also was reported by Hseih et al.⁸  Importantly, vasoocclusive crises eventually abated in all the patients with donor engraftment. A few patients still require narcotics for chronic pain related to their SCD (or as in patient 7 because of an unrelated surgical procedure, correction of a spinal disk herniation).

Allogeneic transplantation remains the only curative therapy for severe hemoglobinopathies.^4,5,8  For patients with devastating complications of this disease, the benefits of transplantation may outweigh the risks if it can be effectively performed.³²  By safely expanding donor eligibility through the use of HLA-haploidentical related donors and high-dose cyclophosphamide posttransplantation, most such patients in need of BMT may now be eligible for this potentially curative therapy. As observed by others,⁸  we confirmed that full donor chimerism is not required to improve symptoms, because patients with mixed chimerism demonstrated improvement in disease-related symptoms such as hemolysis, pain, and transfusion needs. Graft failure remains an obstacle but may be acceptable in a fraction of patients if the majority can be cured with limited serious toxicities.

The authors thank the following individuals for suggestions and comments, as well as for superb patient care and ancillary services delivered to these patients: Dr Richard F. Ambinder; Dr Ivan M. Borrello; Dr George Dover; Dr Douglas Gladstone; Dr Nilanjan Ghosh; Dr Jonathan Gerber; Dr Carol Ann Huff; Dr Yvette L. Kasamon; Dr M. Sue Leffell; Dr William Matsui; Dr Michael A. McDevitt; Dr Alison Moliterno; Dr Jonathan Powell; Dr William Savage; Dr Jerry L. Spivak; Dr Michael Streiff; Dr Lode J. Swinnen; Dr Xuan Yuan; Patrice McMullen, RN; Sandra West; Holly Kemberling, RN; the research staff; the nurses and staff of the Pain Service, Bone Marrow Transplant Coordinator's Office; Bone Marrow Transplant Service; Cell Therapy Laboratory; Pediatric Oncology Unit; and the Hematology Service. The authors also thank Dr Michael DeBaun for critical review of the manuscript.

This research was funded in part by grant P01CA15396 from the National Cancer Institute (R.J.J., principal investigator) and by grant K23HL083089 from the National Institutes of Health (S.M.L.). J.B.-M. is an Investigator-2, Sistema Nacional de Investigadores (Consejo Nacional de Ciencia y Tecnologia, Mexico).

National Institutes of Health

Contribution: J.B.-M., R.J.J., and R.A.B. designed the study, enrolled and cared for patients, analyzed data, wrote the paper, and provided funding; E.J.F. and L.L. helped design the study, cared for patients, analyzed data, and reviewed the paper; S.M.L. enrolled and cared for patients, analyzed data, reviewed the paper, and provided funding; and C.J.G. enrolled and cared for patients, analyzed data, and reviewed the paper.

Conflict-of-interest disclosure: E.J.F., L.L., R.A.B., and R.J.J. are listed as inventors on patent application WO 2009/094456 A3, “Use of high-dose, post-transplantation oxazaphosphorine drugs for reduction of transplant rejection.” S.M.L. served on a scientific advisory board for Hemaquest. The remaining authors declare no competing financial interests.

Correspondence: Javier Bolaños-Meade, MD, Division of Hematologic Malignancies, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Bunting Blaustein Cancer Research Bldg, 1650 Orleans St, Rm 2M-87, Baltimore, MD 21231-1000; e-mail: fbolano2@jhmi.edu.