Main

A central line totally implantable vascular access device (TIVAD) is commonly used in patients with cancer. A TIVAD provides stable, central line access for intravenous chemotherapy, preventing the thrombophlebitis that is common in peripheral veins. In addition, a central line TIVAD facilitates the continuous infusion of some chemotherapeutic agents such as 5-fluorouracil. For patients with advanced or terminal diseases, this device provides vascular access for high-concentration nutritional support.

TIVADs may result in some complications. Intraluminal thrombosis can occur at a frequency of 0.6–0.81 events per 1000 catheter-days (Ray et al, 1996; Schwarz et al, 2000). Manifestations include failure to draw blood from the port or failure of infusion. These events may be reversed using urokinase, streptokinase, or tissue plasminogen activators (Hurtubise et al, 1980; Doizaki and Weber, 1986; Fraschini et al, 1987). Catheter-related venous thrombosis is more severe and can cause swelling of the arm, neck, and head and erythema of the extremities. Five to 41 per cent of patients developed such symptomatic thrombi (Kuter, 2004). Catheter infection is also common and highly associated with intraluminal thrombosis and catheter-related venous thrombosis (Press et al, 1984; Barzaghi et al, 1995; Eastman et al, 2001). All of these complications may delay the chemotherapy course and even be life threatening.

Several factors have been associated with catheter-related thrombotic or occlusion events (intraluminal thrombosis or catheter-related venous thrombosis), such as cancer types, catheter types, and the location of the catheter tips (Stanislav et al, 1987; Anderson et al, 1989; Eastridge and Lefor, 1995; Craft et al, 1996; De Cicco et al, 1997; Nightingale et al, 1997; Schwarz et al, 2000; Kuter, 2004; Shivakumar et al, 2009). Previous studies also reported that compared with catheters inserted on the right side, those inserted on the left side resulted in more catheter-related thrombotic events (Gould et al, 1993; Craft et al, 1996; Shivakumar et al, 2009; Murray et al, 2013). However, no randomised studies have yet investigated the difference between TIVADs inserted on the left and right sides. Because all previous studies were nonrandomised observational or retrospective analyses, their results could be confounded by other risk factors. Therefore, we conducted a randomised observational study to investigate whether the side of central line TIVAD implantation was associated with the occurrence of thrombotic or occlusion events in patients with cancer.

Patients and methods

Study design

This randomised observational study was conducted at National Taiwan University Hospital (NTUH), Taipei, Taiwan. After obtaining informed consent from all patients, they were randomly assigned an implantation side. Randomisation was performed by investigators with no access to the patients by using a free randomisation software developed by Dr M. Saghaei from Isfahan University of Medical Sciences in Iran (http://mahmoodsaghaei.tripod.com/Softwares/randalloc.html). Lung cancer and ovarian cancer were used as stratification factors. According to the standard procedures, the patients received central line TIVAD implantation surgery. This study was approved by the Research Ethic Committee of NTUH and registered at www.clinicaltrials.gov (NCT01525277).

Patient eligibility

Patients were eligible for inclusion if they were aged 20 years or older, had received a diagnosis of cancer, and required central line TIVAD implantation. Patients who had breast cancer, mediastinal tumours exceeding 6 cm in diameter, an Eastern Cooperative Oncology Group performance status of >2, difficulty in receiving standing posteroanterior view chest X-ray examination, or a prior history of TIVAD implantation were excluded. In addition, patients with conditions preventing the randomisation of the implantation side were excluded.

Surgical procedures

Within 7 days before surgery, patients’ haemogram, coagulation time, and blood levels of D-dimer and fibrinogen were examined as routine clinical practice at the NTUH laboratory. We followed a previously reported surgical procedure for TIVAD implantation (Lin et al, 2011). Briefly, all patients were sedated, with spontaneous ventilation maintained during the procedure. The longitudinal axis of the axillary vein was identified using ultrasound. The ultrasound probe was held perpendicular to the skin, and the skin was marked for site preparation. The long axis of the axillary vein was assessed by 2D image and confirmed by pulse wave Doppler. Needle insertion trajectory was also determined. In this scanning plane, we made certain there were no overlapping artery or brachial plexus neural tissues. We used an 18-gauge intravenous catheter for venipuncture. The standard Seldinger technique with a peel-away introducer sheath was used to introduce the guide wire and connect to a 7.8-F polyurethane catheter (Port-A-Cath; SIMS Deltec, Inc., St Paul, MN, USA). The position of the catheter tip was confirmed using chest X-ray films obtained immediately after the completion of the procedure in the post-anaesthesia care unit. All implantations were performed by one of two anaesthesiologists (Dr W-Y Lin and Dr C-P Lin) who specialised in this procedure.

Endpoints

The primary endpoint of this study was the occurrence of catheter-related thrombotic or occlusion events, which included catheter-related venous thrombosis and intraluminal thrombosis. Catheter-related venous thrombosis was diagnosed through vascular duplex ultrasound (non-compressibility of the vein) or computed tomography as routine clinical practice at NTUH when patients experienced symptoms such as swelling and pain of the arm. Intraluminal thrombosis was defined as failure of infusion or failure to withdraw blood from TIVADs. Secondary endpoints included time to the occurrence of catheter-related thrombotic events, catheter-related venous thrombosis, intraluminal thrombosis, and catheter-related infection.

Catheter-related infection included catheter-related bloodstream infection (CRBSI) and catheter colonisation. CRBSI was defined as culture of the same organism from both the catheter segment and at least one blood culture from a peripheral vein or culture of the same organisms from at least two blood samples, one from the peripheral vein and the other from TIVAD, with the latter to be reported as positive at least 2 h sooner than the former (Chatzinikolaou et al, 2004). Catheter colonisation was defined as >15 colony-forming units in culture of the catheter segment or positive blood culture via TIVAD with pathogens typically causing catheter-related infection (Staphylococcus epidermidis, Staphylococcus aureus, and Candida species) (Hentrich et al, 2014).

Follow-up

We combined regular telephone contacts, patient reports, and medical record reviews for follow-up. Telephone contacts were performed every 4 weeks with a medical record review in the meantime to determine the aforementioned endpoints. Follow-up was continued until catheter removal or death.

Maintenance care for all catheters was provided by nurses in the chemotherapy room and oncology-related wards who have received adequate training. No blood withdrawal from TIVADs was permitted unless there was need for blood culture. All TIVADs were flushed with heparin solution 100 U ml−1 after every use or every 6 weeks if they were not utilised for a prolonged period.

Statistical analysis

The patient number was estimated based on the hypothesised thrombotic or occlusion event rates (10% for right-sided implantation and 15% for left-sided implantation), with α=0.05 and β=0.20 under 1 : 1 randomisation. The loss to follow-up rate was expected to be 10%. A total of 246 patients were required. The Kaplan–Meier method was used to estimate survival, and the log-rank test was used in a univariate analysis to compare survival between groups. To estimate the hazard ratios of factors predicting the occurrence of endpoints in univariate and multivariate analyses, a Cox proportional hazards model was used. All statistical analyses were performed using the SAS statistical software (Version 9.4, The SAS Institute, Cary, NC, USA). A two-sided P-value of ⩽0.05 was considered statistically significant.

Results

Patient characteristics

Between 27 March 2012 and 30 March 2015, 246 patients signed informed consent. Of these, six patients withdrew before randomisation, and another five patients withdrew after randomisation but before undergoing the surgical procedure (Figure 1). Finally, 235 patients received the TIVAD implantation according to the protocol.

Figure 1
figure 1

Patient flow diagram.

Full size image

The characteristics of patients in the randomised patient cohort (n=240) and the per-protocol cohort (n=235) are listed in Table 1. All patients received TIVAD implantation for chemotherapy administration. Demographic data were similar between the two cohorts. In the per-protocol cohort, 47% of the patients were female, and the most common cancer type was colorectal cancer (33%), followed by lung (22%), pancreatic (9%), and gastric (9%) cancer. Furthermore, 34% and 44% of the patients had stage III and stage IV cancers, respectively, upon enrolment. The mean age of the patients was 60.7 years. Before the implantation, five patients regularly received antiplatetlet agents; however, no patients received regular anticoagulation. Venous puncture was successful at the first attempt in most (87%) patients; the mean number of venous puncture attempts was 1.3. No accidental arterial puncture occurred during venous puncture.

Table 1 Patient characteristics of the randomisation cohort and the per protocol cohort
Full size table

After randomisation, 117 and 118 patients received TIVAD implantation on the left and right sides, respectively. The patients who received implantation on different sides generally had similar demographic data. However, the men (P=0.042) and the patients with lymphoma (P=0.044) were more likely to receive left-sided implantation (Table 1). Left-sided implantation was associated with a higher possibility of the catheter tip being placed at the right atrium-superior vena cava junction (P=0.001; Table 1).

Patient outcomes

By the last follow-up date of 31 March 2016, the median follow-up time was 24.8 months; 73 (62%) patients with left-sided TIVAD implantation and 73 (62%) patients with right-sided TIVAD implantation had discontinued the study. The most common causes were patient death and catheter removal. The median time to catheter removal or death was 22.8 months (95% confidence interval, 17.7–27.9).

Catheter-related thrombotic or occlusion events occurred in 9 (4%) patients, accounting for 0.065 events per 1000 catheter-days. The occurrence rates were similar between the patients with left- and right-sided implantations (5 vs 3%, P=0.333; Table 2). When we compared time from catheter implantation to the occurrence of these thrombotic or occlusion events, the results between the two groups remained similar (P=0.328, Figure 2A). The 24-month thrombosis or occlusion rate was 5.7% and 3.6% in the patients with left- and right-sided catheter implantation, respectively.

Table 2 Patient outcomes
Full size table
Figure 2
figure 2

Time to ( A ) all catheter-related thrombotic or occlusion events and ( B ) catheter-related infection. Patients are grouped by the side of the TIVAD implantation. P-values are conducted using the log-rank test.

Full size image

Catheter-related venous thrombosis occurred in only 2 (1%) patients, accounting for 0.014 events per 1000 catheter days. These events occurred in one patient with left-sided implantation and one patient with right-sided implantation, so the side of implantation had no significant associations with the occurrence of vessel thrombosis (P=1.000). Both venous thrombosis events improved after anticoagulation, and the TIVADs remained functional. Intraluminal thrombosis occurred in 7 (3%) patients; all of these events were reversed using urokinase or heparin.

Catheter-related infection was observed in 26 (11%) patients (Table 2), accounting for 5.7 events per 1000 catheter-days. The occurrence rate was similar between the patients with left- and right-sided implantations (9 vs 14%, P=0.221), as was the time from catheter implantation to the occurrence of catheter-related infection (P=0.222, Figure 2B). The 12-month catheter-related infection rate was 7.6%, and 9.7% in the patients with left- and right-sided implantations, respectively; the 24-month infection rate was 10.7% and 14.8%, respectively. CRBSI and catheter colonisation occurred in 9% and 5% of the patients, respectively.

Among patients with catheter-related infection, the median time to infection was 8.3 months (Supplementary Figure S1). The most commonly identified pathogens included Staphylococcus epidermidis (n=4) and Staphylococcus aureus (n=3) (Supplementary Table S1). Gram-positive bacteria, gram-negative bacteria, and Candida species were identified in 11 (42%), 10 (38%), and 5 (19%) patients, respectively.

We examined several potential factors associated with catheter-related thrombosis. Age, sex, cancer diagnosis, disease stage, previous mediastinal radiotherapy, tip location, handedness, number of venous puncture attempts, operator, body mass index, and performance status, were analysed first in a univariate and then in a multivariate analysis (Table 3). Platelet count, prothrombin time, activated partial thrombin time (APTT), D-dimer level, and fibrinogen level before surgery were also included. The side of the implantation remained unassociated with catheter-related thrombosis in the multivariate analysis. Moreover, longer APTT was associated with less catheter-related thrombosis (adjusted hazard ratio, 0.653; P=0.050).

Table 3 Univariate and multivariate analysis of potential predictive factors of thrombotic events by a Cox proportional hazards model
Full size table

Discussion

In this prospective randomised observation study, we observed that the side of central line TIVAD implantation was not a predictive factor for catheter-related thrombotic events or infection. Many potential confounding factors, such as catheter types and operators were well controlled. Most other confounding factors were equally distributed in the two randomised groups. Although the unexpected low incidence of catheter-related thrombotic or occlusion events reduced the statistical power of the study, this is the first prospective randomised study to address this issue.

The low incidence of catheter-related thrombotic or occlusion events in this study could be attributed to multiple factors. Almost 40% of patients still had TIVAD in place, so the incidence will further increase. We excluded patients with a very high possibility of catheter-related thrombotic events, such as patients with large mediastinal tumours. We did not perform active surveillance (i.e., regular Doppler ultrasound examinations); thus, all catheter-related venous thrombosis events in this study were symptomatic. However, the prevalence of such symptomatic venous thrombi remained much lower than that reported in Western countries, which was approximately 9–12% (Kuter, 2004; Beckers et al, 2010). All of the patients in this study were of Asian ethnicity, which may be associated with a lower chance of venous thromboembolism (Klatsky et al, 2000; Lee et al, 2006; Oh et al, 2008).

Both of the operators who performed the implantation in this study were well experienced in performing this surgical procedure. The number of attempts in venous puncture, which was reported as a risk factor for catheter-related thrombotic events (Lee et al, 2006; Murray et al, 2013), was generally low in our study; this was probably due to the selection of ultrasound-assisted subclavian insertion (Biffi et al, 2009). Almost 90% of the patients required only one venous puncture, which reduced the endothelial damage. All of the patients in this study received implantation of catheters made of polyurethane, which may also reduce thrombosis compared with catheters made of polyvinylchloride, tetrafluoroethylene, or polyethylene (Linder et al, 1984; Gallieni et al, 2008; Shivakumar et al, 2009).

We found that the patients who received TIVAD implantation on the left side were more likely to have the catheter tip at the optimal location, which was the right atrium-superior vena cava junction (Caers et al, 2004; Kuter, 2004; Verso et al, 2008). Gender distribution was imbalanced in the per-protocol cohort, and the male patients and patients with lymphoma were more likely to receive TIVAD implantation on the left side. However, after including the tip location, gender, and cancer diagnosis as covariates in the multivariate analysis, the side of the implantation was still not identified as a predictive factor for catheter-related thrombotic or occlusion events.

The incidence of catheter-related infection or CRBSI was slightly higher than that reported in the literature (Eastman et al, 2001; Chen et al, 2013; Freire et al, 2013; Wang et al, 2015). In our study, more than 40% of the patients had stage IV disease; 9% of the patients had haematological disease and another 9% of the patients had pancreatic cancer. These factors may have contributed to the higher incidence of catheter-related infection (Toure et al, 2012; Lipitz-Snyderman et al, 2014; Wang et al, 2015).

In conclusion, we demonstrated that the side of central line TIVAD implantation was not associated with the occurrence of catheter-related thrombotic or occlusion events in patients with cancer.