Clinical Predictors of Response to Anti-PD-1 First-Line Treatment in a Single-Centre Patient Cohort: A Real-World Study.

AIMS: Cutaneous melanoma is one of the most immunogenic tumours. Immunotherapy with checkpoint inhibitors, such as anti-PD-1 antibodies, has significantly improved the prognosis in metastatic melanoma. However, only half of the patients respond to this therapy and have a favourable outcome. Identifying factors associated with treatment failure and early identification of responders are both important to select the best treatment approach for each patient. The aim of our study was to investigate clinical biomarkers of response to treatment with anti-PD-1 antibodies. MATERIALS AND METHODS: We selected all patients with stage IV melanoma (n = 147), subjected to first-line treatment with anti-PD-1 in the last 10 years. We investigated the associations between patients' different clinical features and progression-free survival, using the Cox proportional hazards models. RESULTS: In the multivariate analysis, an increased risk of disease progression was observed among patients with stage M1d metastases (hazard ratio 3.30; 95% confidence interval 1.58-6.91), compared with patients with stage M1a-M1b. Moreover, the risk of progression was greater in patients with the Eastern Cooperative Oncology Group Performance Status (ECOG PS) 1 (hazard ratio 2.04; 95% confidence interval 1.02-4.06) and in patients with ECOG PS ≥ 2 (hazard ratio 2.19; 95% confidence interval 1.05-4.55) compared with ECOG PS 0. High levels of lactate dehydrogenase (hazard ratio 2.06; 95% confidence interval 1.18-3.59) and the presence of respiratory diseases (hazard ratio 4.14; 95% confidence interval 1.42-12.0) at the beginning of anti-PD-1 treatment were also associated with an increased risk of disease progression. In a subgroup analysis, neutrophil count and neutrophil/lymphocyte ratio before anti-PD-1 treatment were higher in patients who underwent disease progression. CONCLUSION: In our study population, independent predictors of disease progression among patients treated with first-line anti-PD-1 were as follows: ECOG PS, staging, lactate dehydrogenase and the presence of respiratory diseases.

Expression of connexin57 in mouse development and in harmaline-tremor model.

Connexin57 (Cx57) was previously reported in retinal cells but not in brain nerve cells. This occurrence was tested in this study, by searching for the expression of Cx57 RNA and protein transcripts during the postnatal development of the mouse CNS. Both the Cx57 RNA (investigated by reverse transcriptase-polymerase chain reaction (RT-PCR)) and the protein (Western-Blot and immunohistochemistry using a polyclonal antibody generated in chicken) transcripts were firstly expressed in the late postnatal development (P12). The expression of Cx57 in adult life (studied at P28, by in situ hybridization and immunohistochemical analysis) concerned few regions of the brain stem (inferior olive, lateral reticular nucleus and motor trigeminal nucleus), the cerebellum (Purkinje cells and cerebellar nuclei) and the spinal cord (alpha-motoneurons). Double immunohistochemical studies using the Cx57 antibody and antibodies, which specifically labelled neuronal nuclei (NeuN) and astrocyte cells glial fibrillary acidic protein (GFAP), showed the expression of Cx57 segregated in neuronal cells. The study also confirmed the expression of Cx57 in the horizontal cells of the retinal outer plexiform layer, reported in previous investigations. Given the expression of Cx57 in the cerebellum and pre-cerebellar nuclei, such as olivary and lateral reticular nuclei, a possible role of Cx57 was hypothesized in the electrical coupling of the cerebellum. This hypothesis was tested by searching for the expression of the Cx57 transcripts in the mouse cerebellum of the harmaline-tremor model. The up-regulation of the Cx57 transcripts reported in this model suggested a possible involvement of Cx57 in the electrotonic coupling of the cerebellar system.

Expression of pannexin2 protein in healthy and ischemized brain of adult rats.

The expression pattern of the pannexin2 protein (Px2) in healthy and ischemized brains of adult rats was investigated. A polyclonal antibody for rat Px2 was generated in chicken and purified for affinity. This antibody was used to study by Western blot, Enzyme-Linked Immunosorbent Assay, and immunohistochemistry, the expression pattern of Px2 in healthy brain of adult rats and in the hippocampus of rats submitted to bilateral clamping of carotid arteries for 20 min, followed by different times of reperfusion (I/R) (8 h, 24 h, 48 h, 72 h, 14 days and 30 days). Immunohistochemical studies visualized the wide and complex expression pattern of Px2 in the healthy brain. All Px2(+) positive cells were neurons which also showed no puncta on their cellular membranes. Both pyramidal cells and interneurons, the majority of which were positive to parvalbumin, were stained in healthy hippocampus. The number of Px2 interneurons in the hippocampus showed a progressive reduction at successive time intervals after I/R, with a negative peak of about -40% after 72 h from I/R. Interneurons which were positive for both Px2 and parvalbumin, represented about the 85% of all parvalbumin cells stained in the hippocampus. This percentage rested grossly unmodified at different time intervals after I/R in spite of the progressive neuronal depletion. Concomitantly, an intense astrogliosis occurred in the hippocampus. Most of the astroglial cells expressed de novo and for a transient time (from 24 h to 14 days from I/R), Px2. Primary co-cultures of hippocampal neurons and astrocytes were submitted to transient ischemia-like injury. This set of experiments further confirmed the in vivo results by showing that Px2 is de novo and transiently expressed in astroglial cells following a transient ischemia-like injury. These results suggested the expression of Px2 in the astrocytes may be induced either from injured neurons or by biochemical pathways internal to the astrocyte itself. In conclusion, our results showed the transient expression of Px2 in astrocytes of reactive gliosis occurring in the hippocampus following I/R injury. We hypothesize that Px2 expression in astrocytes following an ischemic insult is principally involved in the formation of hemichannels for the release of signaling molecules devoted to influence the cellular metabolism and the redox status of the surrounding environment.

Expression of pannexin1 in the CNS of adult mouse: cellular localization and effect of 4-aminopyridine-induced seizures.

The expression pattern of pannexin1, a gene coding for a protein that forms gap junction channels, was studied as both mRNA and protein in the CNS of adult mouse. Pannexin1 was widely expressed in the CNS by neuronal cell types but not glial cells, except for Bergmann glial cells of the cerebellar cortex. Cells positive to Ca-binding proteins, principally parvalbumin, but also calbindin and calretinin, as well as glutamate decarboxylase 67 kDa isoform, were pannexin1-positive. Pannexin1 labeling was found in cells which are known to exhibit spontaneous and synchronous discharge, such as neurons of the inferior olivary complex and the reticular thalamic nucleus, and also in neurons whose electrical activity is not coupled with neighboring cells, such as motoneurons of the spinal cord. The analysis of cellular localization showed puncta that surrounded cell bodies (e.g. the pyramidal cells of hippocampus) or restricted areas inside the cell bodies (e.g. the spinal motoneurons). In Bergmann glial cells the staining was present as fine grains that covered a large part of the cellular surface. Pannexin1 stained cells that previous studies have reported as expressing connexin36, another protein forming gap junction channels. Thus, it was possible that these two proteins could be integrated in the same functions. Since connexin36 expression levels change after seizures, we examined the expression of both pannexin1 and connexin36 in cerebral cortex, hippocampus, cerebellum and brain stem at different time intervals (2, 4 and 8 h) after i.p. injection of 4-aminopyridine, which resulted in systemic seizures. The only modification of the expression levels observed in this study concerned the progressive decrement of the connexin36 in the hippocampus, while pannexin1 expression was unchanged. This finding suggested that pannexin1 and connexin36 are involved in different functional roles or that they are expressed in different cell types and that only those expressing the Cx36 are induced to apoptosis by epileptic seizures.

Progenitor cells from the adult mouse brain acquire a neuronal phenotype in response to beta-amyloid.

Neurospheres from adult mouse subventricular zone (SVZ) were grown in suspension cultures for 12-15 days. Neurospheres consisted mainly of neural precursor cells (NPCs) immunoreactive for nestin and also contained nestin-negative precursors. We used these neurospheres to determine the effects of synthetic beta-amyloid fragments (both betaAP(1-42) and betaAP(25-35)) on NPC proliferation, differentiation and survival. We show that neurospheres exposed to 25 microM betaAP(25-35) or betaAP(1-42) for 24 h (a toxic condition for mature neurons) did not undergo apoptosis. Instead, betaAP(25-35) orientated nestin-negative precursors towards nestin-positive NPCs and turned nestin-positive NPCs into neuroblasts. Intracerebroventricular infusion of full-length betaAP(1-42) increased the population of PSA-NCAM-positive cells in the SVZ, without affecting proliferation. We conclude that betaAP influences the fate of progenitor cells, driving their differentiation towards a neuronal lineage.

A phase II study of irinotecan plus chronomodulated oxaliplatin, 5-fluorouracil and folinic acid in advanced colorectal cancer patients.

The combination of irinotecan (CPT-11), oxaliplatin (L-OHP), 5-fluorouracil (5-FU) and folinic acid (FA) is one of the possibilities to overcome chemoresistance in advanced colorectal cancer (ACRC) patients. The aim of this study was to determine the tolerability and activity of CPT-11 plus chronomodulated infusion of L-OHP, 5-FU and FA in ACRC patients. A total of 35 patients (91% pretreated, 77% with CPT-11, 54% with L-OHP, 42% with both) were treated every 3 weeks with CPT-11, 180 mg m(-2) day 1 i.v., plus L-OHP, 20 mg m(-2) day(-1), 5-FU, 700 mg m(-2) day(-1) and FA, 150 mg m(-2) day(-1), all three drugs from day 2 to day 5 by chronomodulated infusion. The patients' (pt) data were as follows: male/female 21/14; median age 58 years (range: 38-70); PS 0: 26 pts (74%), PS 1: 8 pts (23%), PS 2: 1 pt (3%); primary tumour colon/rectum 26/9; involved organs: 1, 14 pts (40%); 2, 17 pts (48%); >or=3: 4 pts (11%); previous chemotherapy lines 1: 12 pts (34%), 2: 10 pts (28%), >or=3: 10 pts (28%). A total of 221 courses (c) were performed; no grade 4 toxicity was observed with only one grade 3 (G3) neutropenia and thrombocytopenia (3%) in one out of 221 courses (<1%). Maximal toxicity (G3) was nausea and diarrhoea in 10 pts (28%), occurring in 14 out of 221 c (6%) and 12 out of 221 c (5%) respectively. Seven patients achieved a partial response (20%, confidence interval (c.i.) 6.8-33.3) and one patient a complete response (2.9%, c.i. 0-8.4), for a total overall response rate of 22.9% (c.i. 9-36.8); 15 out of 35 (42.9%, c.i. 26.5-59.3) had stable disease and 12 out of 35 (34.3%, c.i. 18.6-50) patients underwent a progression. In conclusion, this four-drug regimen is feasible in advanced pretreated ACRC patients with no significant haematological toxicity and acceptable diarrhoea. The activity of this combination is currently studied in EORTC 05011 study.

Multiple zonal projections of the nucleus reticularis tegmenti pontis to the cerebellar cortex of the rat.

Compartmentalization (alternating labelled and unlabelled stripes) of mossy fibre terminals was found in the cerebellar cortex after iontophoretic injections of biotinylated dextran amine into discrete regions of the nucleus reticularis tegmenti pontis (NRTP). The zonal pattern was only observed when volumes of nuclear tissue ranging from 4.5 x 106 to 17.66 x 106 microm3 were impregnated. Up to nine compartments (i.e. up to five stripes separated by four interstripes) were found in crus I and in vermal lobule VI. Up to seven compartments (four stripes and three interstripes) were found in crus II; up to five compartments (three stripes and two interstripes) were identified in the lobulus simplex, the paraflocculus and vermal lobules IV, V and VII; up to three compartments (two stripes and one interstripe) were identified in the paramedian lobule and, finally, up to two compartments (one stripe and one interstripe) were identified in the copula pyramidis, in the flocculus and in vermal lobules II, III, VIII and IX. The projections of the NRTP are arranged according to a divergent/convergent projection pattern. From single injections in the NRTP, projections were traced to a set of cortical stripes widely distributed over the cerebellar cortex. The set of stripes labelled from different regions of the NRTP partially overlapped but complete overlap was never found. This finding revealed that the topographic combination of the projections of the NRTP to the cerebellar cortex is specific for each region of the NRTP. Finally, the projections to single cortical areas were arranged according to a pattern of compartmentalization that is specific for each cortical area, independent of the site of injection in the NRTP and of the number of stripes evident in the cortex.

Laterality of the pontocerebellar projections in the rat.

This study aimed to investigate the trajectory of fibres from the pontine nuclei that reach the two sides of the cerebellum. Injections of biotinylated dextran amine (BDA) were made within the basilar pontine nuclei (BPN) and the nucleus reticularis tegmenti pontis (NRTP) in one side of rats with electrolytic injury of the middle cerebellar peduncle (MCP), ipsilateral or contralateral to the side of injection. Fibres were traced from the pontine nuclei (BPN and NRTP) to both sides of the cerebellum passing through the respective MCPs. The study carried out in rats with injury to one peduncle showed projections segregated to the half-side of the cerebellum innervated by the intact peduncle. The laterality observed was confirmed by a retrograde tracer study. In fact, injections of different fluorescent tracers in rats with injury of single MCP showed that in the pontine nuclei only cell bodies stained by the tracer injected in the half-cerebellum ipsilateral to the intact peduncle. Finally, similar injections (i.e. different fluorescent tracers in symmetric areas of the cerebellar cortex) in the cerebellum of intact brain rats showed that BPN and NRTP differ for the laterality of their projections. In fact, 82% of BPN cells project contralaterally and 18% ipsilaterally, whereas 60% of NRTP cells project contralaterally and 40% ipsilaterally. In conclusion, this study showed that the MCPs receive fibres from the pontine nuclei of both sides and project to the ipsilateral half of the cerebellum and that different contingents of projections to the two sides of the cerebellum arise from BPN and NRTP.

Quality of life and chronotherapy.

The importance of evaluating patient's quality of life (QoL) in clinical practice and research is recognized clearly in oncology. In the advanced phase of disease such an evaluation represents an endpoint as important as survival. Quality of life is both a subjective and multidimensional concept evaluated mainly by validated questionnaires. In colorectal trials involving advanced stage disease the effects of different chemotherapy treatments on QoL were evaluated. Almost all the studies found no deterioration in QoL during chemotherapy. The European Organization for the Research and Treatment of Cancer (EORTC) Chronotherapy Study Group utilized three different approaches to assess QoL. The first centered on the stability of QoL during a 6mon treatment period in patients undergoing chronotherapy. The second centered on research of the biological and clinical determinants of QoL involving features of the circadian activity rhythm and patient survival and the relationship between QoL and patient performance status, response to therapy, and psychosocial variables as well as drug-induced toxicity. The third centered on the clinical effectiveness of psychological intervention on patients undergoing chronotherapy to improve psychosocial status during treatment. This papers reviews the results of EORTC Chronotherapy Group studies on QoL.

Single-agent oxaliplatin in pretreated advanced breast cancer patients: a phase II study.

PURPOSE: Oxaliplatin (L-OHP), a new platinum analogue, is an active drug in colorectal and ovarian cancer. In this phase II study we explored tolerability and activity of oxaliplatin as a single agent in metastatic breast carcinoma patients. PATIENTS AND METHODS: Fourteen anthracycline pretreated advanced breast cancer patients were enrolled. Oxaliplatin was given at 130 mg/m2 on day 1 and repeated every three weeks. Analysis of toxicity, response rate and survival was performed. RESULTS: The median number of courses per patient was four (range 2 6). The median administered dose-intensity was 43.3 mg/m2/week (range 32.5-43.3) which represents 100% of projected dose-intensity. No severe toxicity was encountered. Three patients developed acute transient laryngeal symptoms. Three patients displayed a partial response (21%), (95% confidence interval (CI): 0%-43%), two stable disease (14%) and nine progressed (64%). Response lasted five, four and five months respectively. Median survival was 12 months. CONCLUSIONS: In this limited experience, oxaliplatin appeared to be well tolerated and moderately active in advanced anthracycline-pretreated breast cancer patients. Combination chemotherapy with other active drugs such as 5-fluorouracil (5-FU), anthracyclines and taxanes should represent the next step of development of this new drug.

Corticonuclear projections of the cerebellum preserve both anteroposterior and mediolateral pairing patterns.

The aim of the present study was to establish whether a diverging arrangement of the corticonuclear cerebellar projections exists and, if so, what relation it has with the inferior olivary complex. Iontophoretic injections of a 1 : 1 mixture of tetramethylrhodamine dextran amine and biotinylated dextran amine into the cerebellar cortex orthogradely labelled fibre terminals in the cerebellar nuclei and retrogradely labelled cell bodies in the inferior olivary complex. The injections were into A, B, C2, C3, D1 and D2 bands. These injections showed diverging projections to the cerebellar nuclei, with 'primary projections' directed to the nuclear region previously reported to be specifically connected with the injected band and 'secondary projections' directed to other nuclear regions. Secondary projections from the A, C2 and C3 bands diverged to nuclear regions primarily controlled by cortical bands lateral to those injected. Secondary projections from the D1, and D2 bands diverged to nuclear regions primarily controlled by cortical bands medial to those injected. Moreover, injections distributed along the D1 and D2 bands showed similar sets of nuclear targets, while those distributed along the A, C2 and C3 bands showed two sets of nuclear targets in relation to the anteroposterior location of the injected area within these bands. The cortical areas that projected to the same set of nuclear targets were innervated from single olivary regions, while those that projected to different sets of nuclear targets were innervated from different subsets of single regions of the inferior olive. The results suggest that the olivary bands of the cerebellar cortex project to the cerebellar nuclei with a diverging pattern that is organized in both the mediolateral and the anteroposterior axes.

Multiple zonal projections of the basilar pontine nuclei to the cerebellar cortex of the rat.

This study revealed a sagittal zonal pattern of projections to the cerebellar cortex after hydraulic or iontophoretic injections of anterograde tracers (tritiated leucine, wheat germ agglutinin-horseradish peroxidase, or biotinylated dextrane amine) in the basilar pontine nuclei of Wistar rats. The zonal pattern of projection was observed only after injections of small size, whereas large injections labeled diffusely wide areas of the cerebellar cortex, masking the zonal projection because the fusion of contiguous stripes. Diverging projections to discrete sets of sagittal stripes in the two sides of the cerebellar cortex arose from single injections. The stripes of fiber terminals were sharply delimited on both sides by areas, interstripes, either virtually void of labeling or with a much lower density of labeling. Thus, the areas of the cerebellar cortex were parceled in sets of sagittal compartments, stripes and interstripes, by the pontine projections. Up to five compartments (three stripes and two interstripes) were observed in the paraflocculus, in the copula pyramidis, and in vermal lobule IX. Up to nine compartments (five stripes and four interstripes) were found in the crus I, the lobulus simplex, the paramedian lobule, and vermal lobules VI-VIII. Up to seven compartments (four stripes and three interstripes) were found in the crus II. Single injections into the basilar pontine nuclei usually labeled symmetric areas of the cerebellar cortex, which, in some cases, showed similar number of stripes. When this was not the case, the stripes were usually more numerous in the contralateral than in the ipsilateral side. All areas of the cerebellar cortex were projected upon, with zonation patterns from different regions of the basilar pontine nuclei. The projections of the basilar pontine nuclei to the cerebellar cortex were arranged according to a fixed pattern specific for each cortical area, independently of the number of stripes labeled within. The mean width of the stripes visualized in the single cortical areas of different rats was similar, despite the different size of the injections. The length of the stripes ranged widely in the various areas of different rats. The data collected in this study are consistent with the idea that all the mossy afferents to the cerebellar cortex are arranged with a zonal pattern.

Genotype-phenotype correlations and clinical diagnostic criteria in Wolf-Hirschhorn syndrome.

We report on a clinical-genetic study of 16 Wolf-Hirschhorn syndrome (WHS) patients. Hemizygosity of 4p16.3 was detected by conventional prometaphase chromosome analysis (11 patients) or by molecular probes on apparently normal chromosomes (4 patients). One patient had normal chromosomes without a detectable molecular deletion within the WHS "critical region." In each deleted patient, the deletion was demonstrated to be terminal by fluorescence in situ hybridization (FISH). The proximal breakpoint of the rearrangement was established by prometaphase chromosome analysis in cases with a visible deletion. It was within the 4p16.1 band in six patients, apparently coincident with the distal half of this band in five patients. The extent of each of the four submicroscopic deletions was established by FISH analyses with a set of overlapping cosmid clones spanning the 4p16.3 region. We found ample variations in both the size of the deletions and the position of the respective breakpoints. The precise definition of the cytogenetic defect permitted an analysis of the genotype-phenotype correlations in WHS, leading to the proposal of a set of minimal diagnostic criteria, which in turn may facilitate the selection of critical patients in the search for the gene(s) responsible for this disorder. We observed that genotype-phenotype correlations in WHS mostly depend on the size of the deletion, a deletion of <3.5 Mb resulting in a mild phenotype, in which malformations are absent. The absence of a detectable molecular deletion is still consistent with a WHS diagnosis. Based on these observations a "minimal" WHS phenotype was inferred, the clinical manifestations of which are restricted to the typical facial appearance, mild mental and growth retardation, and congenital hypotonia.

Expression of connexin36 mRNA in adult rodent brain.

A new member of the connexin gene family, named Connexin36 (Cx36) has, recently, been identified in rodents and shown to be preferentially, if not exclusively, expressed in neurones of the adult CNS. In this study we present a detailed in situ hybridization analysis of the expression pattern of mouse Connexin 36 (mCx36) mRNA in the adult mouse brain, with particular regards to the correlation of mCx36 expression to specific neuronal cell classes and systems. We found that mCx36 was strongly and widely expressed in the brain, including areas where the presence of gap junctions was never detected before. Quantitative analysis of the hybridization signal indicated varying levels of expression in different areas. In particular mCx36 was highly expressed in the neurones at different levels of the motor pathway, the olfactory pathway, the hippocampus, and areas related to the generation of respiratory rhythm. On the contrary, mCx36 was more heterogeneously expressed in nuclei of the sensory pathways. These findings show that mCx36 is the first connexin specifically expressed in neuronal cells in the adult rodent brain. The profiles of expression clearly indicate that mCx36 might play specific roles within different neuronal systems.

Ifosfamide given by continuous-intravenous infusion in association with vinorelbine in patients with anthracycline-resistant metastatic breast cancer: A phase I-II clinical trial.

BACKGROUND: Vinorelbine (VNR) is highly active in metastatic breast cancer (MBC) and has shown an overall response rate of 40%-50% as first-line treatment. In vitro, a synergy has been observed between this drug and ifosfamide (IFX). In addition, the pharmacokinetics of IFX suggest that it may have greater activity when given by continuous-intravenous infusion (C.I.V.I.). The aim of this study was therefore to assess the antitumor efficacy and toxicity of the combination of bolus VNR and C.I.V.I. IFX as second-line therapy in anthracycline-resistant breast cancer patients. PATIENTS AND METHODS: Forty-two patients with MBC who had already received anthracycline-based chemotherapy were treated with a regimen consisting of IFX, by C.I.V.I. for 72 hours and bolus VNR. The courses were repeated every three weeks for a maximum of eight cycles. Four dose intensification steps were planned. IFX, 1.5 g/m(2) on days 1-3 + VNR, 30 mg/m(2) on day 1 (six patients); IFX, 2 g/m(2) on days 1-3 + VNR, 25 mg/m(2) on day 1 (six patients); IFX, 1.8 g/m(2) on days 1-3 + VNR, 25 mg/m(2) on days 1 and 8 (six patients); IFX, 2 g/m(2) on days 1-3 + VNR, 25 mg/m(2) on days 1 and 8 (24 patients). Sodium-2-mercaptoethane sulfonate (mesna) was associated with IFX at an infusion ratio of 1:1 and, once the infusion was completed, per os every four hours for three times. RESULTS: All of the 42 patients entered were assessable for toxicity, and 41 of them for response. Neutropenia was the most frequently-occurring toxicity, but only five patients at the highest dose level (11.9%) presented grade 4, and none of those at the first three steps. Other significant toxic effects were mild (only grade I-II). The median relative dose intensity was 95% at the highest dose level and all the treatments were administered on an out-patient basis. The overall response rate was 36.5% with a CR rate of 4.8% (two of 41 patients, all at the highest dose level) and a PR rate of 31.7% (13 of 41 patients). The median response duration was 7.0 months (range 2-13 months). CONCLUSIONS: The present phase I-II study shows that the IFX and VNR combination is an active and well-tolerated treatment in MBC and provides an alternative to taxanes for patients previously treated with anthracyclines.

Carboplatin, ifosfamide, and vinorelbine in the treatment of advanced non-small-cell lung cancer: a phase II study.

The authors evaluated the efficacy and toxicity of the combination of carboplatin, ifosfamide, and vinorelbine in the treatment of advanced non-small-cell lung cancer. From March 1994 through March 1996, 56 previously untreated patients with stage IIIB or stage IV non-small-cell lung cancer with measurable lesions and good performance status were entered in the study. The chemotherapy schedule was as follows: carboplatin 100 mg/m2 and ifosfamide 1,500 mg/m2 with mesna on days 1, 2, and 3; vinorelbine 25 mg/m2 on days 1 and 8, every 21 days; for a total of six courses. Among 55 evaluable patients there were three complete responses (5%) and 22 partial responses (40%), for a response rate of 45% (95% confidence interval, 32-59%). The median response duration was 10.3 months (range, 2.5-27.7 months), and median survival time was 11.3 months (range, 1.1-28.1 months). The survival rate at 1 year was 48%. Toxicity included hematologic toxicity in 60% of the 247 treatment cycles administered, nausea, alopecia, and neuropathy. One pathologic complete response was observed in a patient with stage IIIB disease who became operable after four courses of chemotherapy. The outpatient treatment with carboplatin, ifosfamide, and vinorelbine shows activity in advanced non-small-cell lung cancer. The toxicity was well tolerated by patients with a good performance status.