Review by: Rei Mitsuyama, Brown University, 2020


            Hematologic malignancies are a heterogeneous group of cancers for which cytotoxic chemotherapy is the mainstay of treatment. Patients often require ports (implantable venous access systems) to maintain reliable venous access for drug administration. Despite advances in antimicrobial therapy and dedication to infection control, infection remains one of the most common long-term complications of ports. Furthermore, it is the most common reason for premature port removal [1].

            The US Centers for Disease Control describes two types of port infections – port-site and bloodstream infection (BSI) [2]. Port-site infections are defined clinically by superficial erythema of the skin overlying the port or along the subcutaneous tract regardless of blood culture results. Diagnosis of BSI requires at least one of the following: (i) recognized pathogen on at least one blood culture that is not related to an infection at another site, (ii) at least one of the following signs or symptoms: fever (> 38℃), chills, or hypotension, (iii) signs and symptoms of positive laboratory results not related to an infection at another site, and (iv) common commensal cultured from more than two blood cultures drawn on separate occasions. Together, port-site infection and BSI have an incidence of 0.15-0.43 per 1,000 port-days. Early port infection is defined by the Society of Interventional Radiology as infection occurring within the first 30 days of port placement.

            Patients with hematologic malignancies have higher reported rates of port infections compared to patients with solid malignancies. This is likely due to a combination of impaired immunity and longer duration of chemotherapy. Because patients with hematologic malignancies are immunocompromised, the impact of port infections can be substantial. These infections can lead to ICU admissions, longer hospital stays, increased medical expenses, and delays in chemotherapy administration.

            To date, there are no guidelines to screen for patients who are at high risk for port infections. This single-center retrospective study aimed to identify clinical predictors for port infections in adult patients with hematologic malignancies [3].

Study Population and Treatment Protocol

          223 adult patients with hematologic malignancies underwent port placement between January 2012 and December 2015.  Mean age at time of port insertion was 55 years (range, 19–88 y). The most common diagnoses were Non-Hodgkin lymphoma (56.1%), Hodgkin lymphoma (17.0%), and acute lymphocytic leukemia (8.1%).

            All ports were placed by the Division of Interventional Radiology with nearly half (47.1%) of the ports were placed in an inpatient setting. All coagulopathies were corrected before port placement and port placement was avoided in those with an absolute neutrophil count (ANC) <500 due to reported risk of port infection [4]. All patients were given antibiotic prophylaxis with cefazolin or clindamycin for penicillin allergies. The choice of single-lumen vs. double-lumen port was made at the discretion of the referring oncologist. The right internal jugular vein was accessed under ultrasound guidance. The left internal jugular was used if the right side was thrombosed or had been previously irradiated.

Data Collection

            All ports were followed by the referring oncology service. The date and type of port infections along with rate of infections per 1000 catheter-days was determined by chart review. Patients were excluded from analysis if they had death unrelated to port infections, loss of follow-up, or still had a port in use. Predictors for early and overall port infections were identified with proportional subdistribution hazard regression (PSHREG) analysis. PSHREG analysis is a method for studying survival data, which analyzes duration of time until an event of interest, in our case, port infection. PSHREG can be used to determine how specific variables such as medications, laboratory values, among others, could affect the probability of said outcome.

Relevant medications included for analysis were steroids, anticoagulants, and antiplatelet agents. Laboratory values at the time of port placement were included when analyzing early port infections but not when looking at survival incidence of overall port infections. Variables with P < 0.1 in univariate, looking at one variable at a time, PSHREG analysis were then included in backward stepwise multivariate, multiple variables, PSHREG analysis. For each variable, the estimated hazard ratios with 95% confidence intervals were reported. To compare time to port infection for the significant variables on multivariate analysis, cumulative incidence function (CIF) was estimated and compared with a Gray test of equality. P < 0.05 was considered statistically significant.


            Total duration of follow-up was 83,722 catheter-days (median 274). Eight patients (3.6%) were found to have early port infections giving a 30-day port infection rate of 1.2 per 1,000 catheter-days. Five patients had port infection within the first 15 days. Seven of the patients with early port infections had BSIs and 3 early port infections had port-site infections. The ports were removed in all patients at the time of early port infections. No patients with early port infections were taking relevant medications. Hypoalbuminemia was the only variable that was statistically significant in multivariate analysis (hazard ratio[1]  = 5.03; 95% confidence interval, 1.14–22.16; P = 0.03). Patients with hypoalbuminemia at the time of port placement had a statistically higher CIF than those without (P = 0.02).

            After the first 15 days, 26 patients (11.7%) developed port infections, with an overall rate of 0.3 infections per 1,000 catheter-days. 22 of these patients (84.6%) had BSIs, 5 of whom also had port-site infections. Ports were removed in 24 patients while the other 2 patients were treated with oral antibiotics. In univariate analysis, steroid use, double lumen catheter, and diagnosis of lymphoma, AML, or leukemia reached overall statistical significance. Following multivariate analysis, only steroid use remained statistically significant (hazard ratio = 3.41; 95% confidence interval, 1.55–7.47; P = 0.002) and the CIF for patients on steroids was statistically greater than patients not on steroids (P = 0.01).


            Cytotoxic chemotherapy is the mainstay of treatment for hematologic malignancies and is often delivered via ports. Despite our best efforts, infection remains one of the most common and significant long-term complications of using theses ports.

Currently, no guidelines exist to screen for patients who may be at high risk for port infections. Defining high-risk variables may allow us to intervene and potentially reverse parameters that predispose patients to infection. Past studies have suggested variables such as neutropenia, double lumen catheter use, and outpatient placement of ports as risk factors. This study only identified hypoalbuminemia as an independent risk factor for early port infections and steroid use as an independent risk factor for overall port infections. Hypoalbuminemia is a sign of a protein-deficient state and may be remedied by improved nutrition. Unfortunately, steroids are an integral part of many chemotherapy regimens and therefore this is likely not a variable that can be readily changed.

The retrospective nature and small sample size of this study led to several limitations. While patients with any hematologic malignancy were included, the majority were patients with non-Hodgkin lymphoma. Thus, these results may not be generalizable to those with other malignancies. Furthermore, the large number of variables analyzed relative to the small sample size limited the statistical power of this study. Both of these major limitations could likely be addressed with a larger, multicenter study.


  1. Fischer L, Knebel P, Schroder S, et al. Reasons for explantation of totally implantable access ports: a multivariate analysis of 385 consecutive patients. Ann Surg Oncol 2008; 15:1124–1129.
  2. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008; 36:309–332.
  3. Zhang, Shunqing et al. Clinical Predictors of Port Infections in Adult Patients with Hematologic Malignancies. Journal of Vascular and Interventional Radiology, Volume 29, Issue 8, 1148 – 1155
  4. I. C. Chen, C. Hsu, Y. C. Chen, S. F. Chien, H. F. Kao, S. Y. Chang, F. C. Hu, K. H. Yeh; Predictors of bloodstream infection associated with permanently implantable venous port in solid cancer patients, Annals of Oncology, Volume 24, Issue 2, 1 February 2013, Pages 463–468
  5. The PHREG Procedure
  6. Dignam, James J and Maria N Kocherginsky. “Choice and interpretation of statistical tests used when competing risks are present” Journal of Clinical Oncology .Vol. 26,24 (2008): 4027-34.