ALBERT

Description: Background: Prolongation of severe neutropenia is known to predict the onset of pneumonia and invasive fungal infection in patients undergoing chemotherapy. The D-index is a promising tool to assess the severity of neutropenia, and the utility of the D-index is being investigated in clinical research. However, there are a few reports regarding the association between the D-index and the onset of opportunistic infection. We undertook this study to examine whether the D-index is useful to predict the onset of various infections in febrile neutropenia (FN) patients. Methods: The D-index was originally developed by Portugal et al. It is calculated as the area between

Description: Purpose: Stomatitis in allogenic hematopoietic stem cell transplant (HSCT) recipients sometimes causes serious complications. Stomatitis causes disturbances in oral functioning thus diminishing quality of life (QOL) in transplant patients; these include such simple tasks such as feeding, swallowing and speaking. However, given this serious complication, little information is available regarding the prophylactic effect of oral intervention in hematopoietic stem cell transplantation (HSCT). Methods: We retrospectively analyzed the incidence and severity of stomatitis after allogeneic HSCT with or without oral intervention among 96 consecutive patients in our hospital between January 1988 and March 2006. Inclusion criteria were conditioning regarding both conventional (CST) and reduced-intensity regimens (RIST), donor source that included both bone marrow and peripheral blood omitting cord blood, HLA disparity eligible to one or two locus mismatched related or unrelated donor. We used a combination following two strategies; one is cryotherapy which has applied since 2003, and another is oral health care which has been used since 2004. The incidence and severity of stomatitis was evaluated during the start of chemotherapy until day 100 after transplantation. Stomatitis was evaluated according to NCI CTCAE v3.0. ‘Cryotherapy’ is defined as the use of cold to treat an injury. Generally, an ice tip is used to make the mucosa cold. ‘Oral health care’ is defined as a prevention or treatment, including that administered by oneself, encompassing any of the immunologic, sensory, neuromuscular and structural functions of the mouth and craniofacial complex. Results: The mean age was 42.7 years. Among 96 patients, 41 patients were treated by CST and 55 by RIST. The incidence of stomatitis was 30.9% (17/55) in RIST, which was significantly lower than the 90.2% (37/41) in CST (P 〈 0.001). Among these 96 patients, severe stomatitis (grade 3 to 4) was observed in 19 (46.3%) CST cases and in 6 (10.9%) RIST cases (P

Description: Abstract 999 Human CD1d-restricted invariant natural killer T (iNKT) cells are a unique subset of innate T-cells that carry an invariant T-cell receptor (TCR) Vα24 chain paired with a TCR Vβ11 chain. They constitutively express a variety of activation and memory markers and possess the capacity to rapidly produce a variety of cytokines including IFN-γ and IL-4 upon TCR engagement. As an immune response launches, iNKT cells can induce innate immune responses and serve as a bridge between innate and adaptive immunity. In addition, it has also been shown that iNKT cells themselves have anti-tumor and anti-infectious effects in mice. However, to translate these findings and develop iNKT cell-mediated immunotherapy, iNKT cells must recognize target cells while sparing normal cells. Autoreactivity is primarily dictated by 1) the specificity and avidity of the TCR on iNKT cells; 2) the expression level of CD1d on target cells; and 3) the density and stability of endogenous ligand(s) presented by CD1d on target cells. In this study, we sought to determine the TCR Vβ11 CDR3 sequence motifs that distinguish high avidity and low avidity iNKT cells. We developed an artificial antigen-presenting cell (aAPC) to expand iNKT cells by transducing K562 with CD1d, CD80, and CD83. Using this CD1d+aAPC loaded with or without α-galactosylceramide (αGC), we expanded CD1d-restricted iNKT cells by stimulating primary CD3+ T cells expressing canonical iNKT TCR Vα24 chain. Expanded iNKT cells were stained with αGC-loaded CD1d tetramers and were found to express TCR Vβ11 chain in conjunction with transgenic canonical TCR Vα24 chain. Sequence analysis of cloned Vβ11 CDR3 revealed that the clonality of iNKT cells generated using unloaded aAPC was significantly lower than that of iNKT cells generated using loaded aAPC. Surprisingly, when cotransfected with canonical TCR Vα24 chain, some TCR Vβ11 chains isolated from iNKT cells generated using unloaded aAPC were stained with unloaded CD1d tetramers produced in HEK293 cells. This result suggests that these reconstituted iNKT TCR recognized endogenous ligand(s) derived from HEK293 cells in the context of CD1d. A comprehensive analysis of the structural avidity demonstrated that these TCR Vβ11 chains reconstituted TCR with significantly higher structural avidity than those cloned from NKT cells generated using loaded aAPC. However, this significant difference was only observed when the structural avidity was measured using unloaded tetramers but not αGC-loaded tetramers. A univariate analysis found that structural avidity was significantly higher when 1) Vβ11 CDR3 used J2-5; 2) Vβ11 CDR3 consisted of exactly 23 amino acids; and 3) Vβ11 CDR3 encoded 3 or more acidic amino acids (asparagic and glutamic acids). A multivariate analysis confirmed that all three variables were independent predictors of higher structural avidity. Furthermore, each variable is sufficient to increase the structural avidity of an iNKT TCR with low structural avidity. Intriguingly, unloaded mouse CD1d tetramers produced in HEK293 cells also stained a human iNKT TCR with high structural avidity reconstituted on both human and mouse T cells. This result suggests that the human iNKT TCR with high structural avidity possessed a cross-species reactivity to mouse CD1d in conjunction with endogenous ligand(s) derived from HEK293 cells. We next studied the autoreactivity of cloned human iNKT TCR with high structural avidity. A human T cell line, Jurkat, reconstituted with high avidity iNKT TCR, was able to recognize cells expressing CD1d endogenously (itself, SUP-T1, and primary human monocytes) and ectopically (K562, C1R, Hela). Furthermore, mouse T cells expressing human iNKT TCR with high structural avidity recognized mouse cell lines, B16, EL4, and 58, all endogenously expressing CD1d. Finally, human primary T cells transduced with the iNKT TCR with high structural avidity were autoreactive, secreting both IFN-γ and IL-4 in response to CD1d+ target cells. Using our CD1d+ aAPC-based system, we successfully isolated human iNKT cells with high structural avidity and identified the TCR Vβ11 CDR3 sequence motifs that dictate the structural avidity of iNKT cells. To develop clinically effective iNKT cell-mediated immunotherapy without unwanted autoimmunity, it will be critically important to define the range of iNKT TCR avidity that enables reactivity to pathologic but not normal cells. Disclosures: No relevant conflicts of interest to declare.

Description: Background: Comorbidity may influence the treatment and long-term quality of life of chronic myeloid leukemia (CML) patients treated with tyrosine kinase inhibitors (TKIs). The Charlson Comorbidity Index (CCI) assesses comorbidity by taking into account the severity of 19 precedent comorbid conditions [J Chron Dis 1987; 40: 373]. The CCI was originally introduced as a measure of the mortality risk of general hospitalized patients, and it can estimate the risk of morbidity in a patient with any medical background. The CCI has been validated to predict patients' life expectancy in various underlying diseases, and it is widely used for both hospitalized patients and outpatients. We evaluated the risk of comorbidity at the diagnosis of CML to determine whether treatment with tyrosine kinase inhibitors (TKIs) results in longer survival in a cohort of Japanese patients. Patients and method: We used a chart review to survey CML patients diagnosed between November 2001 and December 2012 in Kagawa, Japan, where we collected all of the patients based on their registration in another population-based study. Inclusion criteria were (1) diagnosis of CML in the chronic phase during the study period and (2) treatment by TKIs at any point of the study period. Patients of all ages were included. The available TKIs were imatinib, nilotinib and dasatinib. We did not exclude patients treated with two or more TKIs. We used the CCI to evaluate concomitant underlying disease at diagnosis, and we calculated the Sokal and Hasford scores to compare the predictability of prognosis with the CCI at diagnosis. We used the Sokal and Hasford scores as the standard reference. Results: Eighty (47 male, 33 female) cases were enrolled (median age 56 yrs, range 6-89 yrs). The distribution of CCI scores at diagnosis were 2-11 (median 2). The initial treatment was started in 73 cases by imatinib, two cases by nilotinib, and four by dasatinib (one is unknown). Only one patient underwent stem cell transplantation after 2-yr treatment by imatinib, because of the development of myelodysplastic syndrome (this case was not censored.) As of the last follow-up, the treatment responses were 46 major molecular responses (MMRs), 12 complete cytogenetic responses (CCyRs) and 14 complete hematological responses (CHRs). Seventy-five percent of the cases (60/80) achieved CCyR at 12 mos after TKI administration. We observed only five deaths during the 55.5-mos median follow-up period (0.3-217 mos; 3 pneumonia, 1 cancer and 1 unknown cause). The number of patients according to the given CCI comorbidity risk categories are presented in Table 1. The patient numbers in the risk categories (low/intermediate/high) per the Sokal and Hasford scores were 33/27/7 and 21/43/3, respectively. The CCI scores were significantly correlated with the Sokal and Hasford scores (R 〉0.5). Twenty-seven cases with a CCI score ≥3 (52 cases had CCI

Description: Background: Central venous catheters (CVCs) are necessary for critically ill patients requiring intravenous pharmacological intervention and subsequent parenteral nutritional support. Although CVCs allow delivery of medications and nutritional support that cannot be administered safely through central venous, their use is inevitably associated with adverse events, mechanical complication and catheter-related infection. While ultrasound guide has already been proven to decrease mechanical complications, it is not fully elucidated whether ultrasound guide decrease the risk of catheter-related infection. Methods: We observed consecutive CVC insertions between April 2009 and January 2013. In total, 395 insertion cases were surveyed in the hematological oncology unit. We divided the research period into two terms: before December 2011 (early term) and after January 2012 (latter term). Between the early and latter terms, there were substantial differences regarding the use of ultrasound guides. Because insertion maneuvers changed from blind to ultrasound-guided approach after 2012. SMAC Plus MicroNeedle (15G, 13 cm or 12G, 20 cm; Covidien Tokyo, Japan) was used. Practitioners determined which CVC device and which insertion site was preferred for each patient. To determine the clinical efficacy of chlorhexidine gluconate dressing (CHGD) at catheter insertion site, we performed matched cohort analysis among the patients who underwent stem cell transplantation. Total 44 cases were included in the cohort from the total study population. Results: Underlying diseases included hematological malignancies and immunological disorders such as auto-immune diseases and solid organ malignancies. A total of 235 and 160 cases were included in early and latter terms, respectively. Insertion duration was a median 26 days (range, 2-126 days) in the early term and 18 days (range, 2-104) in the latter term. During the early term, the insertion sites were 22.6%, 40.2%, and 25.7% at the cervical, subclavian, and femoral veins, respectively, and 32.3%, 16.9%, and 25.4% in the latter term, respectively. The ultrasound-guided insertion method became a routine practice in the latter term. The frequency of catheter-related blood stream infection (CRBSI) was 8.46/person-days and 13.62/person-days in the early and latter terms, respectively. Mechanical comorbidities decreased from 0.39 (13/33) incidences/month to 0.00 (0/13) after the introduction of the ultrasound-guided insertion method. Using subgroup analysis, detected causative pathogens of CRBSI did not differ between the two terms: gram-positive coccus, gram-positive bacillus, and gram-negative bacillus were 68.9%, 11.5%, and 14.8% in the early term and 68.2%, 11.4%, and 18.2% in the latter term, respectively. Controlled-cohort study revealed CHGD decreased CRBSI caused by Staphylococcus spp. significantly, but not overall other organisms. Conclusion: Ultrasound-guided insertion did not decrease the incidence of CRBSI. However, in our survey, insertion through the cervical vein approach clearly increased after the introduction of the ultrasound-guided method, and mechanical complications decreased significantly. Ultrasound guide and cervical vein approach trended for patients with hematological diseases who required CVC and those who concurrently harbored multiple risks, including thrombocytopenia. The prevalence of CRBSI slightly increased with this trend. With respect to CRBSI, the preventative effect of CHGD in patients undergoing allogeneic stem cell transplantation was promising. We advocate trending of ultrasound-guided, cervical approach CVC with CHGD can provide the safest parental alimentation for the patients with hematological malignancies. Disclosures No relevant conflicts of interest to declare.

Description: Abstract 782 The goal of cellular immunotherapy is to build long-lasting anti-tumor immunologic “memory” in patients and reject tumors for a lifetime. Previously, we and others demonstrated that IL-15 promotes the generation of T cells with a central memory (CM) phenotype which have the capacity to persist and establish effective anti-tumor memory in vivo. Furthermore, it has been shown that CD83 delivers a CD80-dependent T cell stimulatory signal that allows T cells to be long-lived. Based on these findings, we developed a system to generate large numbers of long-lived antigen-specific CD8+ T cells with a memory phenotype. This in vitro culture system utilizes IL-15 and a standardized, renewable artificial antigen presenting cell (aAPC) which was produced by transducing CD80, CD83, and HLA-A*0201 to the human cell line, K562. This aAPC can uniquely support the priming and prolonged expansion of large numbers of antigen-specific CD8+ CTL which display a central/effector memory (CM/EM) phenotype, possess potent effector function, and can be maintained in vitro for 〉1 year without any feeder cells or cloning. We hypothesized that adoptive transfer of these CTL with a CM/EM phenotype should result in anti-tumor memory in humans even without lymphodepletion or high dose IL-2. For our “first-in-human” clinical study, we chose the melanoma antigen MART1 as a target antigen, since MART1-specific HLA-A*0201+-restricted precursor CTL are detectable in some melanoma patients and can be immunophenotyped pre-infusion. Autologous CD8+ T cells were stimulated weekly with peptide-pulsed human cell-based aAPC and expanded with low dose IL-2 and IL-15. After three weeks, polyclonal MART1 CTL were reinfused without additional lymphodepletion, chemotherapy, IL-2, or vaccination. Eight study participants have enrolled and received a total of 15 MART1 CTL infusions (31% MART1 multimer positivity, median). All but one subject received two reinfusions where the 2nd graft was produced from CD8+ T cells harvested two weeks after the 1st reinfusion. To date, ≥2×109 CTL with potent effector function and a CM/EM phenotype were successfully generated for all subjects. No dose limiting toxicities were observed at either Dose Level 1 (2×108/m2) or Dose Level 2 (2×109/m2). Clinical activity was observed with a response by RECIST criteria in 1 subject, which was confirmed by a negative PET/CT 100 days following the last CTL infusion. In addition, 1 patient experienced a mixed response, 1 had stable disease, 3 had progression, and 2 are currently on active therapy. Multimer staining showed that, immediately post infusion, the percentage of CD8+ T cells specific for MART1 temporarily increased in all subjects, with the highest (6.5%) observed in subject #7. In 4 subjects, sustained increases in the frequency of MART1 specific T cells by more than two-fold (range 2.0-10x) for ≥21 days were observed despite the fact that no exogenous cytokines or vaccination was administered. Moreover, an increase of detectable MART1 specific T cells which display a CM phenotype was observed in all evaluable subjects and was observed for ≥35 days in 6 of 8 subjects. In subject #2, the conversion of MART1 CTL immunophenotype from a naïve to a mixture of naïve/memory phenotypes was observed for more than 6 months. We identified 10 individual MART1 T cell clonotypes from peripheral CD45RA- memory T cells on day 21. Clonotypic TCR Vbeta CDR3 analysis revealed that CTL grafts contained 7 out of 10 of these clonotypes. Furthermore, 6 clonotypes persisted in the peripheral CD45RA- memory fraction on days 39, 67 and/or 132. In Subject #3, who showed a mixed clinical response, 5 individual MART1 T cell clonotypes were isolated from lung metastases. 4 out of 5 clones were included in the CTL grafts. This finding supports the possibility that infused CTL can traffic and localize to sites of disease. Intriguingly, in both subjects, we were able to identify MART1 CTL clonotypes that were not detectable in the CTL grafts but possibly emerged after CTL infusion, indicating that adoptive transfer of MART1-specific CTL may provoke a de novo antitumor response. Taken together, these results suggest that CM/EM MART1 CTL generated ex vivo using our cell-based artificial APC in the presence of IL-15 may persist in vivo and induce de novo anti-tumor responses. Further enhancement of anti-tumor activity may be achieved through vaccination, cytokine administration, and/or removal of cytokine sinks and inhibitory factors following appropriate lymphodepletion. Disclosures: No relevant conflicts of interest to declare.

Description: Background: Voriconazole (VRCZ) is a novel azole that exerts strong antifungal activity, and it can be administered orally or intravenously. Intra- and inter-individual variability of the VRCZ concentration in metabolism is reported. VRCZ shows nonlinear pharmacokinetics due to its capacity-limited elimination, and its pharmacokinetics are thus dependent on the administered dose. In these circumstances, drug interactions should also be considered. Hematologists commonly administer corticosteroid treatment to patients with hematological diseases, expecting an anticancer effect in some cases and an anti-inflammatory property in other cases. Against this background, we analyzed the influence of corticosteroid use on the serum VRCZ concentration using a population pharmacokinetics (PK) method. Patients and method: This was a single-institute PK study at our hospital. We collected immunocompromised cases treated with VRCZ and with or without corticosteroid from June 2010 to August 2012 (2 yrs + 2 mos). The patients had hematological malignancies or hematological diseases such as thrombocytopenia. The inclusion criteria were: age 〉20 yrs, treated with corticosteroids for a hematological condition, and treated with VRCZ under a diagnosis of possible or probable fungal infection according to the IDSA criteria. VRCZ was predisposed intravenously and consequently administered orally according to its bioavailability. VRCZ was administered 2´/day for both the oral and intravenous routes. The dose was calculated according to the pharmaceutical company's manual. A double dose was used on the initial day as the loading dose. All patients concurrently treated with a corticosteroid were included in the study regardless of the administration route. We defined 'concurrent use of corticosteroid treatment' as only within 3 days before and after the initiation of VRCZ. Each corticosteroid dose was converted to a prednisolone-based dosage according to the anti-inflammatory effect (relative glucocorticoid activity). We used the total corticosteroid dose during VRCZ treatment to evaluate the contributing factors. The serum VRCZ concentration was measured at the trough point by high-performance liquid chromatography. The VRCZ concentration was measured 〉4 days from the initiation of VRCZ. Samples were collected just before the administration in the morning as a trough level. Results: Fourteen cases (4 males, 10 females; median age 56 yrs, range 35-82 yrs) were investigated. The underlying diseases were acute myeloid leukemia (n=3), thrombotic thrombocytopenic purpura (n=3), chronic lymphoid leukemia (n=1), malignant lymphoma (n=1), and other immunocompromised status including autoimmune diseases (n=6). Five patients (35.7%) received chemotherapy; 12 received corticosteroid therapy. The average and median doses of corticosteroid per day were 55.7 and 80.0 mg/wk, respectively (range 0-377). A total of 27 samples were collected. The VRCZ dose ranged from 200 to 600 mg/day, administered orally for 10 patients and intravenously for four at the point of dose monitoring. The median VRCZ concentration was 2.93 mg/mL (avg. 3.00 mg/mL, 0.66-8.83 mg/mL). For the total samples as a whole, the VRCZ concentration was inversely correlated with the corticosteroid dose (r = −0.20). Conversely, 1/VRCZ concentration was weakly correlated with the corticosteroid dose (r = 0.17). We extracted two cases from whom more than four samples could be collected. In both cases, the relation between the VRCZ concentration and corticosteroid dose more clearly illustrated the inverse proportionality, with the correlation coefficients r = −0.68 and −0.77. Laboratory data evaluated simultaneously with the VRCZ concentration measurement showed no concomitance of severe liver damage; the median values of T.Bil, AST and ALT were 0.5 mg/dL (0.2-2.6), 31 IU/L (8-82) and 26 IU/l (3-82), respectively. Discussion: Our PK study suggested that the VRCZ trough concentration is inversely correlated with corticosteroid usage in immunocompromised patients. VRCZ might be metabolized by an enzyme induced by corticosteroid in each individual. Although the inter-individual variance remains to be determined, we suggest that the VRCZ dosage be adjusted closely under therapeutic drug monitoring in cases involving corticosteroid therapy. Disclosures No relevant conflicts of interest to declare.

Description: Abstract 3022 Poster Board II-998 Dendritic cells (DC) are “professional” antigen-presenting cells (APC) that can prime T cells. Their characteristic morphology and phenotype segregate them from other APC. Many studies suggest that mature DC are able to induce potent antitumor T cell immunity that can reject tumors. Based on this, numerous cancer vaccine trials using ex vivo generated DC have been conducted in humans. However, the observed objective response rates in these studies have been disappointing. This could partially be attributed to difficulties in generating large numbers of clinical grade, optimally matured DC. Also, it is widely accepted that the quality and quantity of DC generated ex vivo vary substantially among individuals. We have hypothesized that the generation of standardized artificial DC (aDC) will overcome the time, expense, and suboptimal reproducibility of DC cultures and prompt the development of DC-based immunotherapy for cancer. Previously, we developed a renewable and standardized artificial APC (aAPC) by transducing HLA-A2, CD80, and CD83 to the human erythroleukemic suspension cell line, K562. This aAPC can naturally process and present HLA-A2-restricted peptides and uniquely support the priming and prolonged expansion of large numbers of antigen-specific CD8+ CTL. Generated antigen-specific CTL display a central ∼ effector memory phenotype consistent with in vivo persistence, possess potent effector function, and specifically recognize tumor cells. Furthermore, CTL can be maintained in vitro for a prolonged period of time up to 〉1 year without any feeder cells or cloning. Recent clinical trials have demonstrated that adoptive transfer of anti-tumor CTL with a memory phenotype generated ex vivo using this aAPC, IL-2, and IL-15 can persist in cancer patients as memory T cells for 〉6 months without any lymphodepletion, adjuvants, or cytokine administration. Clinical responses have also been observed in some patients. To develop a standardized aDC, we have undertaken an approach to differentiate our K562-based aAPC into aDC. In neurogenesis, it has been well established that a family of Rho GTPases (Rho, Rac, and Cdc42) critically regulates the outgrowth of neurites, i.e. dendrites and axon. We have found that the inhibition of Rho kinase (ROCK), which is a key effector molecule of Rho, can promote the differentiation of monocyte-derived immature DC into mature DC both morphologically and phenotypically. Intriguingly, when aAPC were forced to attach via a newly identified surface molecule, PladX, and ROCK activity was subsequently blocked, K562-derived aAPC “differentiated” into DC-like cells by acquiring dendrite extensions and growth cone-like structures at the end of the extensions (see picture). PladX-mediated strong attachment was critical for differentiation, since ROCK inhibition without attachment or following attachment via conventional adhesion molecules such as poly-L lysine, fibronectin, or collagen was not sufficient to induce dendrites. Confocal microscopy analysis revealed that dendrites were composed of F-actin rich filopodia and lamellipodia. Furthermore, F-actin and microtubules were differentially localized in the “growth cones” and “dendrite shafts” of aDC, respectively. While treatment with actin inhibitors blocked the generation of “growth cones” but not dendritic shafts, exposure to microtubule inhibitors abrogated the extension of dendritic shafts. Finally, we were able to demonstrate that aDC were more potent than aAPC in CD8+ T cell stimulatory activity. This was the case despite the fact that differentiation of aAPC into aDC does not alter the expression level of molecularly engineered immunoaccessory molecules MHC class I, CD80, and CD83. The effects of the differentiation on processing and presentation of antigenic peptides were negligible since CD8+ T cell antigen was exogenously pulsed as a fully processed synthetic peptide. Taken together, this result indicates that the dendrite formation and the resultant enlarged surface area are critical determinants of DC's enhanced immunogenicity. We have succeeded in producing infinite number of aDC with enhanced immunogenicity by differentiating our renewable and standardized K562-based aAPC, which has been already tested in the clinic. This novel aDC may overcome the cumbersome issues inherent to conventional DC and widen the applicability of DC-based immunotherapy for cancer. Disclosures No relevant conflicts of interest to declare.

Description: Purpose For cancer-bearing patients, especially for patients with hematological malignancies, blood access is a ‘lifeline’ during chemotherapy, in three senses: administration of chemotherapy for the cancer treatment, intravenous supply of nutrients when a patient’s oral intake is decreased, and injection of many agents for supportive care including antibiotics and G-CSF. For these purposes, we usually use a central venous catheter (CVC) at the subclavian portion. Dressing and skin care of the CVC are critical factors influencing the incidence of catheter-related blood stream infection (CRBSI). To clarify the association between preventative procedures and CRBSI rates, we summarize the effectiveness of annually instituted interventions for the prevention of CRBSI and present the result of surveillance of catheter infection in our hospital for a decade, from 1998 to the present. Method This is a prospective cohort study analyzing patients seen in our hospital for the treatment of cancer and predicting neutropenia by observation of catheter infection. All of the patients underwent CVC (Microneedle Seldinger Kit, Safe Guide II, Argyle) insertion with a subclavian approach except for patients with subclavian venous troubles such as embolism or occlusion. Each year for the first five years, we instituted new precautions for preventing CRBSI. Those interventions were introducing low-irritant skin tape from 1999, applying the maximal barrier precaution (MBP) procedure from 2000, applying closed injection-line system (Interlink system; Japan Beckton Dickinson, Tokyo, Japan) from 2001, once-a-week dressing using adherent transparent film (Tegaderm; Sumitomo 3M, Tokyo, Japan) from 2002, and usage of one pair of glove in each procedure for an individual patient from 2003. More than 2 sets (One set means blood samples from peripheral blood and CVC) of blood culture samples were drawn when a patient’s body temperature (axillary) increased to more than 38.0 ºC. Multiple detection of the same isolate in the same individual during a series of febrile episode was considered as one infectious event. Bacteremia was defined as the isolation of at least one pathogen from at least one blood sample. BSI was defined as the recognition of a pathogen from one or more blood cultures that is not related to an infection at another site. CRBSI was defined as bacteremia in a patient with CVC with at least one positive pathogen obtained from a blood culture, and no apparent source for the BSI. We evaluated the duration of CVC insertion, episodes of febrile event, isolated bacteria and bacteremia. Results A total of 55,469 catheter-days were observed. The incidence of CRBSI per one thousand catheter-days was 8.08±2.76 before our interventions (before 1998), 4.18±2.59 after the low-irritant skin tape, 2.84±1.26 after the MBP procedure, 2.18±1.69 after the closed injection-line system, 2.84±2.17 after once-a-week dressing, and 2.53±2.63 after the individual-gloves intervention. After the introduction of MBP, the incidence of CRBSI was significantly decreased from before the start of interventions (P