Identification of cervical intraepithelial neoplasia (CIN) using UV-excited fluorescence and diffuse-reflectance tissue spectroscopy.

BACKGROUND AND OBJECTIVE: The diagnostic potentials of ultraviolet-excitation fluorescence spectroscopy and diffuse-reflectance spectroscopy of tissue are assessed in a study to identify cervical intraepithelial neoplasia (CIN) in vivo. A multivariate algorithm is used to classify tissue into normal tissues, CIN I, and CIN II/III categories, based on spectral characteristics of biopsied tissue sites. STUDY DESIGN/MATERIALS AND METHODS: An optical instrument with the capability of measuring fluorescence and diffuse-reflectance spectra from 120 locations uniformly distributed over the surface of the cervix is described. Using this device, these optical spectra of the cervix were measured on women referred for colposcopy due to an abnormal Pap smear. RESULTS: UV fluorescence differentiates CIN II/III lesions from normal squamous tissue with a sensitivity and specificity of 91 and 93%, respectively. CIN I is distinguished from normal tissue with a sensitivity of 86% and a specificity of 87%. CONCLUSION: Optical spectroscopy shows promise for the detection of pre-cancerous cervical lesions in vivo. The fluorescence and reflectance methods are complementary in their ability to differentiate different tissue types, making the use of the two techniques together more diagnostic than the use of either method separately.

Relative risk of high grade squamous intraepithelial lesion associated with prior abnormal Pap smears.

OBJECTIVE: Pap smear frequency remains controversial, especially for women with consecutive negative smears. We undertook the current study to ascertain the association of high grade squamous intraepithelial lesions (HSIL) and prior abnormal Paps. STUDY DESIGN: Women with biopsy-proven HSIL (cervical intraepithelial neoplasia 2 and 3) diagnosed between September 1996 and December 1997 and age-matched controls with a negative Pap obtained during the same time period were selected. RESULTS: Sixty-three cases (mean age = 32 years) of HSIL and 69 controls (mean age = 33 years) constituted the study population. Any prior abnormal diagnosis conferred a 15-fold increased risk of HSIL on the current Pap (50/63 vs. 14/69, P < .0001). When limited to the 60 women with at least three prior Paps, the odds ratio for HSIL on the current Pap with any prior abnormal was 18 (28/31 vs. 10/29, P < .0001). Three cases had at least three consecutive negative Paps prior to the diagnosis of HSIL. CONCLUSION: Women with one or more prior negative Pap smears had a significantly decreased risk of HSIL on the current Pap. Consecutive negative Paps did not appear to further decrease the risk; 10% of HSIL patients had had three or more consecutive prior negative Paps. To detect HSIL at its earliest stage, women should be advised to continue annual Pap screening in spite of consecutive negative results.

Occult ovarian tumors in women with BRCA1 or BRCA2 mutations undergoing prophylactic oophorectomy.

PURPOSE: To review the findings at prophylactic oophorectomy of a series of women who presented to a familial breast and ovarian cancer clinic. MATERIALS AND METHODS: Data from medical charts, operative notes, and pathology reports were collected on women who had undergone prophylactic oophorectomies because of the elevated risk of ovarian cancer. Because only a subset of patients underwent BRCA1 and BRCA2 testing, each patient's risk of hereditary predisposition was calculated using the Berry-Parmigiani model and family history data. RESULTS: From June 1989 to December 1998, 50 women seen at our clinic underwent prophylactic oophorectomy, 33 of whom had a calculated risk of carrying a germline BRCA1 or BRCA2 mutation greater than 25%. Among this group, four incidental tumors were found (four of 33, or 12%); one tumor was noted at the time of surgery and three were noted only in the final pathology. Two patients had microscopic, poorly differentiated serous adenocarcinomas in multiple sites on both ovaries. A third patient had a bilateral serous borderline tumor with micropapillary features. The fourth patient had a microscopic serous borderline ovarian tumor. All four patients had germline BRCA1 or BRCA2 mutations, and three had unremarkable transvaginal ultrasonography examinations within 6 months before prophylactic surgery. CONCLUSION: Foci of malignant tumor are not uncommon in prophylactic oophorectomies performed in women at very high risk for ovarian cancer and may not be detected on ultrasonograms. Surgeons should have a high suspicion of finding cancer in these women at the time of prophylactic surgery, and careful pathologic assessment of the specimens should be conducted.

Phase I trial of carboplatin, paclitaxel, etoposide, and cyclophosphamide with granulocyte colony stimulating factor as first-line therapy for patients with advanced epithelial ovarian cancer.

OBJECTIVES: The goal of this study was to determine the maximally tolerated doses (MTDs) of carboplatin, paclitaxel (Taxol), etoposide, and cyclophosphamide (CTEC) with granulocyte-colony stimulating factor (G-CSF, Filgrastim) support as first-line chemotherapy in women with advanced epithelial ovarian cancer (EOC). METHODS: Newly diagnosed patients with either stage IV EOC, or stage III EOC and any amount of gross residual tumor after surgical debulking were eligible to receive six cycles of CTEC over five different dose levels in this phase I trial (planned 21-day cycle length). Paclitaxel, carboplatin, and cyclophosphamide were administered intravenously on Day 1, and oral etoposide was administered on Days 1, 2, and 3. G-CSF was administered beginning Day 4. RESULTS: Twenty patients received a total of 98 cycles of CTEC over the five dose levels evaluated. Bone marrow suppression was the major toxic effect, with grade 4 neutropenia and thrombocytopenia being observed in 25 and 23% of cycles, respectively. The overall incidence of febrile neutropenia was 10%, and no toxic deaths occurred. No grade IV thrombocytopenia or febrile neutropenia was observed once the carboplatin dose was reduced from AUC of 7 to 5. Nonhematologic toxicity was generally mild (grade 2 or less). Dose-limiting toxicity was not observed at the highest dose level evaluated in this study, preventing assignment of the MTD. The clinical complete response rate was 92%, although 15 of 16 evaluable patients have progressed with a median progression-free interval of 4 months (range, 2-11 months). One patient remains disease-free 9 months from the completion of CTEC. CONCLUSIONS: The CTEC regimen is well tolerated and highly active. Although the MTD was not reached in this study, the short median progression-free interval suggests that this regimen is unlikely to be superior to standard treatment with paclitaxel and carboplatin. Strategies to optimize the development of future combination chemotherapy regimens in the treatment of newly diagnosed ovarian cancer are discussed.

Early discharge after abdominal surgery: experience on a gynecologic oncology service.

OBJECTIVES: The goals of this study were to determine the length of stay (LOS) after abdominal surgery following implementation of practice guidelines on a gynecologic oncology service, to identify adverse outcomes of early discharge, and to identify clinical predictors of longer LOS. METHODS: A retrospective chart review of 266 consecutive patients who had elective abdominal surgery was performed. Clinical data, LOS, and follow-up data were abstracted. Univariate and multivariate analyses were performed to identify clinical variables predictive of LOS. RESULTS: Mean LOS was 2.94 days. Seven (2.6%) patients were readmitted after discharge. With multivariate analysis, extensive surgical procedures, coronary artery disease, and bowel surgery were predictive of longer LOS (P < 0.05). CONCLUSIONS: Early discharge following abdominal surgery was possible for most patients and was associated with a low rate of readmission. Extensive surgical procedures, coronary artery disease, and bowel surgery were predictive of longer LOS.

Outpatient minilaparotomy for ovarian cysts.

OBJECTIVE: To design an operative procedure for the ambulatory management of ovarian cysts using classical surgical techniques. STUDY DESIGN: One hundred consecutive patients 55 years old or younger with 115 persistent or complex ovarian cysts less than 10 cm in diameter were managed as outpatients by minilaparotomy. Minilaparotomy is defined as a transverse or vertical incision 3-5 cm in length. The procedure and anesthetic were dictated by each clinical situation. Bupivacaine HCl with epinephrine was injected in the wound preemptively, and ketorolac was administered systemically perioperatively. Operative times, complications and pathology were determined for each case. RESULTS: The procedures (unilateral cystectomy, 65; bilateral cystectomy, 9; unilateral salpingo-oophorectomy, 20; and bilateral salpingo-oophorectomy, 6) were performed under general endotracheal anesthesia in 89, laryngeal mask anesthesia in 5 and spinal block in 6. Mean operative time was 46 minutes. Estimated blood loss in 96% of cases was < 50 mL, and none was > 100 mL. Pathology in two cases revealed adenocarcinoma of borderline malignancy. Remaining histology included endometrioma, 40; dermoid, 25; serous cystadenomas, 14; hemorrhagic corpus luteum, 9; mucinous cystadenoma, 8; cystadenofibroma, 7; follicular cyst, 3; fibrothecoma, 2; and peritoneal inclusion cyst, 1. Ninety-six of 100 patients were discharged on the day of surgery. Two were admitted for urinary retention, one for severe nausea and vomiting, and one for diabetes control. CONCLUSION: Minilaparotomy is a safe, rapid procedure for the management of ovarian cysts on an ambulatory basis. It can be performed under regional anesthesia, avoids intraperitoneal spill and requires only basic operative techniques and instrumentation. Minilaparotomy is also a cost-effective technique for outpatient management of ovarian cysts.

A population-based study of BRCA1 and BRCA2 mutations in Jewish women with epithelial ovarian cancer.

OBJECTIVE: To determine the frequency of BRCA1 mutations 185delAG and 5382insC and BRCA2 mutation 6174delT in Jewish women with ovarian cancer and in matched controls in a population-based study. METHODS: Forty-eight Jewish women with epithelial ovarian cancer (32 invasive and 16 borderline) and 33 Jewish control subjects were obtained from a population-based, case-control study of ovarian cancer in eastern Massachusetts and New Hampshire. Mutational analysis on exons 2 and 20 of BRCA1 and exon 11 of BRCA2 was conducted on blood samples from patients and controls. RESULTS: Fourteen (44%) of 32 Jewish patients with invasive epithelial ovarian cancer carried either a 185delAG mutation of BRCA1 (n = 8) or a 6174delT mutation on BRCA2 (n = 6). Neither of these mutations was identified in 16 women with borderline ovarian tumors or in 33 controls. No 5382insC mutation of BRCA1 was identified in any of the patients or control subjects in the series. Family history did not predict mutation status. CONCLUSION: BRCA1 185delAG and BRCA2 6174delT mutations are frequent in Jewish women with invasive epithelial ovarian cancer, irrespective of family history. Genetic counseling might be warranted in women with invasive epithelial ovarian cancer based on Jewish ethnicity alone.

Incidence and predictors of cervical dysplasia in patients with minimally abnormal Papanicolaou smears.

OBJECTIVE: To estimate the incidence of dysplasia in patients with Papanicolaou smears showing atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intraepithelial lesions (SIL) and to identify clinical predictors of dysplasia in these patients. METHODS: Patients referred for ASCUS and low-grade SIL were reviewed retrospectively. All patients were evaluated with immediate colposcopy. A multivariate logistic regression analysis was performed to identify clinical predictors of histologic SIL and histologic high-grade SIL. RESULTS: One hundred thirty-seven (34%) of 406 consecutive patients had histologic SIL. Regression analysis identified age (under 35 versus 35 years or above) and initial smear (low-grade SIL versus ASCUS) as statistically significant predictors of histologic SIL and high-grade SIL (P < .001). When patient outcomes were analyzed by age and initial Papanicolaou smear results, the subgroup of patients 35 years or older with ASCUS had low incidences of histologic SIL (14%) and high-grade SIL (1%). The other subgroups (under 35 years with ASCUS, under 35 years with low-grade SIL, and 35 years or older with low-grade SIL) had incidences of histologic SIL and histologic high-grade SIL of at least 28% and 14%, respectively. CONCLUSION: The high incidence of dysplasia in patients with minimally abnormal Papanicolaou smears suggests that immediate colposcopy might be appropriate for many of these patients. Age and initial Papanicolaou smear are predictive of dysplasia and might be used to select patients who have low incidence of dysplasia and might not require immediate colposcopy.

Absence of constitutive EGF receptor activation in ovarian cancer cell lines.

Previous investigators have noted that certain ovarian cancer cell lines secrete and respond to transforming growth factor-alpha (TGF-alpha), suggesting that endogenous activation of the epidermal growth factor (EGF) receptor through autocrine or paracrine mechanisms might contribute to the proliferative response. In order to determine whether autocrine stimulation was partly responsible for the proliferative response in ovarian cancer, we investigated whether the EGF receptor expressed by ovarian cancer cell lines was constitutively activated as assessed by the presence of tyrosine phosphorylation. A specific anti-phosphotyrosine antibody was used in conjunction with an immunoblotting technique in order to detect EGF receptor phosphorylation in ovarian cancer cell lines in the absence and presence of exogenous EGF. The effects of neutralising anti-EGF receptor antibody on the proliferation of ovarian cancer cell lines was also examined. We found no evidence for constitutive tyrosine phosphorylation of the p170 EGF receptor in eight epithelial ovarian cancer cell lines tested, although each line demonstrated inducible phosphorylation in response to exogenous EGF. The absence of constitutive EGF receptor activation was also noted when cells were grown under high density conditions, thus excluding a role for membrane-bound EGF or TGF-alpha in this process. Media conditioned by five ovarian cancer cell lines, as well as malignant ascites obtained from 12 different ovarian cancer patients, were not capable of stimulating EGF receptor phosphorylation. Finally, the proliferation of ovarian cancer cell lines was not significantly inhibited in the presence of neutralising anti-EGF receptor antibody. These data suggest that EGF receptor activation through autocrine pathways is not a major mechanism for the growth of many ovarian cancer cell lines. Other pathways of signal transduction which bypass the requirement for EGF receptor activation may be important in the proliferation for ovarian cancer cells. Such EGF receptor-independent pathways may limit the effectiveness of strategies designed to inhibit ovarian cancer cell growth through disruption of EGF receptor function.

Strong expression of vascular permeability factor (vascular endothelial growth factor) and its receptors in ovarian borderline and malignant neoplasms.

Angiogenesis is a critical factor in the growth, progression, and metastatic spread of solid tumors. Furthermore, angiogenesis has been correlated with prognosis in patients with ovarian cancer. The pathogenesis of the angiogenic events in ovarian cancer, however, are not well defined. Vascular permeability factor/vascular endothelial growth factor (VPF/VEGF) is a multifunctional cytokine that has been shown to be an important regulator of tumor angiogenesis. The purpose of the present study was to define the expression of VPF/VEGF and its receptors flt-1 and KDR in ovarian tumors. Four specimens of normal ovarian cortex and 41 specimens of benign (4), borderline (8), and malignant (29) ovarian tumors were studied by in situ hybridization, and in some cases by immunohistochemical analysis. VPF/VEGF protein was also determined by an immunofluorometric assay in cyst fluids obtained from 11 patients, including 7 benign, 2 borderline, and 2 malignant tumors. VPF/VEGF mRNA and protein were expressed by the neoplastic cells in all of the malignant tumors evaluated, with the majority of tumors (28 of 29) showing strong expression of mRNA. Serous borderline tumors had variable VPF/VEGF mRNA expression, with two of six cases showing focal strong expression and four showing low-level expression. No definite expression of VPF/VEGF was seen in two cases of mucinous borderline tumors. No strong expression of VPF/VEGF mRNA was observed in normal ovarian cortex, including surface epithelium, or benign tumors. Substantially higher VPF protein concentrations were detected in cyst fluids of the two malignant (60, 440 pM) and two borderline tumors (210, 590 pM) than in the seven benign serous cysts (mean, 10 +/- 3 pM). In addition, microvascular endothelial cells strongly expressed mRNA of the VPF/VEGF receptors flt-1 and KDR and immunostained for VPF/VEGF protein in the majority of malignant and borderline tumors examined. These findings suggest that VPF/VEGF plays an important role in the angiogenesis associated with ovarian neoplasms.

Establishing an interdisciplinary patient care team: collaboration at the bedside and beyond.

The authors describe how an interdisciplinary team used skills in communication and collaboration to improve patient care on a busy surgical service. A major goal was to establish and maintain continuity of care in the face of decreasing lengths of stay and increasing patient acuity. The authors share their insight about designing and supporting a successful interdisciplinary patient care team and discuss how their experiences relate to concepts such as case management and career development.

Estrogen receptor expression is a common feature of ovarian borderline tumors.

PURPOSE: The presence of estrogen receptor (ER) and its therapeutic significance in ovarian borderline tumors (OBT) have not been established. We recently observed a response to tamoxifen therapy given empirically to a patient with unresectable, recurrent serous borderline tumor (SBT). In view of this observation the present study was undertaken to assess ER expression in 51 cases of OBT. MATERIALS AND METHODS: ER expression was determined retrospectively, using an immunohistochemical method on formalin-fixed, paraffin-embedded specimens, from 35 cases of SBTs, 6 cases of mucinous mullerian (MMBT), and 10 cases of mucinous intestinal borderline tumors (MIBT). ER was considered positive if > 5% of tumor epithelial cell nuclei were immunostained. Both SBTs and mucinous borderline tumors (MBTs) were included to determine the influence of histologic type on ER expression. RESULTS: The patients ranged in age from 25 to 77 years (median 43 years for SBTs, 36 years for MMBTs, and 37 years for MIBTs). The stage distribution for the SBTs was stage I in 27 patients (77%), stage II in 4 patients (11.5%), and stage III in 4 patients (11.5%). All patients with MBTs were stage I. ER expression was observed in the majority of cases and correlated with histologic type: 94% (33/35) of SBTs and 100% (6/6) of MMBTs were ER positive compared to 0% (0/10) of MIBTs (P < 0.01). In the SBT category the presence of ER did not correlate significantly with stage or age. In addition, ER was positive in all four SBT implants (including one involved lymph node) and two recurrent SBTs analyzed. CONCLUSION: ER expression is a common feature of SBT and MMBT, but not MIBT. The relevance of ER expression in the pathogenesis and treatment of OBTs requires further investigation.

CD44 variant expression is a common feature of epithelial ovarian cancer: lack of association with standard prognostic factors.

PURPOSE: CD44 is a hyaluronic acid receptor that exists as a standard 90-kd form (CD44S) as well as several CD44 variant isoforms produced through alternative splicing. Expression of CD44 variants is associated with clinically aggressive behavior in some human tumors. The purpose of the present study is to define the expression of CD44 variant isoforms in ovarian cancer and to investigate whether the expression of these molecules is associated with adverse prognosis. MATERIALS AND METHODS: Six specimens of normal ovarian surface epithelium (NOSE) and 31 separate cases of newly diagnosed ovarian cancer were studied by a combination of reverse-transcription polymerase chain reaction (RT-PCR) and immunoperoxidase staining. Clinical correlation was made between CD44 variant expression and stage (I/II v III/IV), residual disease (< or = 2.0- v > 2.0-cm mass), age (< or = 65 v > 65 years), histology (papillary serous v other), grade, and survival. RESULTS: RT-PCR analysis revealed that NOSE predominantly expressed transcripts for CD44S, as well as a restricted pattern of transcripts characteristic of CD44 splice variants. CD44S and CD44 variant exon nine sequences (CD44-9v) were focally expressed in one of two NOSE specimens examined by immunoperoxidase staining. In comparison, the majority (71%) of ovarian cancer specimens expressed a complex pattern of CD44 splice variants by RT-PCR analysis. Immunoperoxidase studies revealed that the majority of ovarian cancer specimens expressed both CD44S and CD44-9v, whereas expression of sequences from variant exons 3, 4, and 6 was uncommon. There was no association between CD44 variant expression (transcript or protein) and stage, residual disease, age, histology, grade, or survival. CONCLUSION: Expression of CD44S and CD44-9v is a common feature of epithelial ovarian cancer cells. The lack of a significant association between CD44 variant expression and prognosis suggests that other factors may be more important in determining the clinical behavior of this disease.

Expression and function of beta 1 and alpha v beta 3 integrins in ovarian cancer.

Ovarian cancer cells disseminate by implanting onto the peritoneal mesothelial cell surface of the abdominal cavity. A common feature of these peritoneal implants is the presence of tumor cell invasion into the submesothelial extracellular matrix (ECM). In view of the important role of integrins in ECM recognition and cell migration, we were interested in defining the pattern of integrin expression and function in ovarian cancer cell lines and primary tissue samples. The beta 1 integrin chain was expressed by CAOV-3, SKOV-3, OVCAR-3, and SW626 ovarian cancer cell lines, associated with expression of alpha 1, -2, -3, -5, and -6 chains. The alpha 4 chain was also expressed by two of the four lines. In addition to beta 1 integrins, the alpha v beta 3 integrin was also expressed, although there was no expression of beta 2, -4, and -7 chains. Immunoprecipitation of surface-labeled CAOV-3 cells with an anti-beta 1 antibody revealed a band at approximately 110-130 kDa consistent with the known molecular mass of the beta 1 chain, as well as several associated bands consistent with noncovalently linked integrin alpha chains. A similar pattern of beta 1 and alpha v beta 3 integrin expression was observed for primary ovarian cancer tissue samples. Ovarian cancer cell lines exhibited significant binding to collagen type I and laminin which was primarily mediated by beta 1 integrins. In contrast, ovarian cancer cell binding to fibronectin was mediated by both alpha 5 beta 1 and alpha v beta 3 integrins. Even though mesothelial cells were observed to express fibronectin mRNA and protein, binding of ovarian cancer cells to peritoneal mesothelium was not blocked by neutralizing antibodies to beta 1 or alpha v beta 3 integrins. These data suggest that functional integrins are commonly expressed by ovarian cancer cells, although they do not appear to mediate ovarian cancer cell implantation onto peritoneal mesothelium. The role that integrins play in the invasion of ovarian cancer cells into the submesothelial ECM deserves further investigation.

Functional heterogeneity of CD44 molecules in ovarian cancer cell lines.

We have previously shown that CD44 partly mediates ovarian cancer cell attachment to peritoneal mesothelium through recognition of mesothelial-associated hyaluronate. CD44 is a major receptor for hyaluronate and exists as a standard 90-180-kDa form (CD44H), as well as several higher molecular mass variant forms produced by alternative splicing. To determine whether functional differences exist between CD44H and its variants we have investigated the relationship between CD44 isoform expression and mesothelial adhesion in 12 ovarian cancer cell lines. Eight lines were CD44 positive (range, 83-94%) and demonstrated significant binding to mesothelium and hyaluronate, whereas two lines showed reduced CD44 levels (3-13%) and demonstrated decreased binding. Interestingly, two other lines (OVC-3 and SW626) expressed CD44 in the majority of cells (>93%) and yet bound weakly to mesothelium. Mean linear fluorescence intensity of CD44 expressed by OVC-3 and SW626 cells was approximately one-half that of strongly binding cell lines, suggesting that the ability to adhere may be partly related to CD44 surface density. However, immunoprecipitation and immunoblot analyses revealed that standard CD44H represented only 23-31% of total CD44 in weakly binding cells, with the majority of species being comprised of CD44 variants. Indirect immunofluorescence of OVC-3 and SW626 cells confirmed the presence of CD44 variants containing exons v3, v6, and v9. In contrast, CD44H represented the majority (75-86%) of total CD44 expressed by strongly binding cell lines such as CAOV-3 and UPN36T. Transfection of CD44H cDNA into weakly binding OVC-3 cells restored significant mesothelial binding which was partly blocked by anti-CD44 antibody. These data suggest that the expression of CD44 is necessary but not sufficient for mediating attachment of ovarian cancer cells to mesothelium. Although CD44 variants may constitute the major CD44 species in certain ovarian cancer cell lines, it appears that these CD44 species are not always capable of mediating significant binding to mesothelium or hyaluronate. Rather, an adequate level of CD44H is the critical determinant of binding in this system. The role of CD44 variants in the process of ovarian cancer metastasis will require further investigation.

Use of peripheral-blood progenitor cells abrogates the myelotoxicity of repetitive outpatient high-dose carboplatin and cyclophosphamide chemotherapy.

PURPOSE: Attempts to increase dose-intensity in clinical practice have been limited by cumulative hematologic toxicity despite the use of hematopoietic growth factors. To address this problem, we designed a study to determine whether four cycles of dose-intensive chemotherapy with carboplatin could be administered in the outpatient setting using granulocyte-macrophage colony-stimulating factor (GM-CSF) and peripheral-blood progenitor cells (PBPCs) that had been harvested before initiation of treatment. PATIENTS AND METHODS: An initial cycle (cycle no. 0) of cyclophosphamide 4 g/m2 followed by GM-CSF was used to mobilize PBPCs harvested by leukapheresis for 6 consecutive days. Cycles no. 1 through 4 consisted of outpatient carboplatin 600 mg/m2 and cyclophosphamide 600 mg/m2 followed by GM-CSF 5 micrograms/kg subcutaneously (SC) twice per day every 28 days. In cycle no. 1, PBPC were not reinfused to assess the effects of GM-CSF alone. In cycles no. 2 through 4, PBPCs were reinfused on day 3 in an outpatient setting. RESULTS: In eight assessable patients, the addition of PBPCs in cycle no. 2 resulted in a significant reduction in the median duration of thrombocytopenia less than 20,000/microL (6.5 v 1 day; P = .016), days to platelets more than 50,000/microL (20.5 v 15 days; P = .020), number of platelet transfusions (five v 1.5; P = .016), and duration of neutropenia (absolute neutrophil count [ANC] < 1,000/microL (7 v 2.5 days; P = .008) when compared with cycle no. 1. Dose-limiting hematologic toxicity, defined as more than 7 days of platelets less than 20,000/microL or ANC less than 500/microL, was observed in four of eight patients during cycle no. 1, but not during cycles no. 2, 3, and 4 of chemotherapy supported by PBPCs (a total of 19 cycles in eight patients). Five of eight patients completed all four cycles of high-dose therapy. Three patients did not complete four cycles due to late thrombocytopenia (n = 2) or tumor progression (n = 1). CONCLUSION: These results indicate a benefit of PBPCs in addition to GM-CSF in alleviating myelosuppression of dose-intensive chemotherapy. Initial collection of PBPCs may allow administration of repetitive cycles of high-dose chemotherapy with acceptable toxicity to outpatients at disease onset.