Schematic illustration of the chemisorption of perfluoro-alkylsiloxane chains on 1 side of a silicone rubber voice prosthesis. During argon-plasma treatment, 1 side of the prosthesis was shielded from the plasma by a plaster cast, and only the argon-plasma–treated side was immersed in the silane solution.
Scanning electron micrograph of a partially surface-modified Groningen button voice prosthesis removed from a tracheoesophageal shunt after 4 weeks of use. The right side of the prosthesis was modified by chemisorption of long perfluoro-alkylsiloxane chains (8 fluorocarbon units). The bar represents 1.0 mm.
Everaert EPJM, Mahieu HF, van de Belt-Gritter B, Peeters AJGE, Verkerke GJ, van der Mei HC, Busscher HJ. Biofilm Formation In Vivo on Perfluoro-Alkylsiloxane–Modified Voice Prostheses. Arch Otolaryngol Head Neck Surg. 1999;125(12):1329-1332. doi:10.1001/archotol.125.12.1329
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
To study the influence of perfluoro-alkylsiloxane (PA) surface modification of silicone rubber voice prostheses on biofouling.
Placebo-controlled clinical trial.
Tertiary referral center, with specialization in head and neck cancer treatment.
Eighteen consecutive patients with laryngectomies and experienced in the use of a voice prosthesis who visited the outpatient clinic for prosthesis replacement.
Eighteen partially surface-modified voice prostheses (3 with short-chain PAs [1 fluorocarbon unit] and 15 with long-chain PAs [8 fluorocarbon units]) were inserted via the patients' tracheoesophageal shunts and remained in place for 2 to 8 weeks.
Replacement of the prostheses.
Main Outcome Measures
Evaluation of biofilm formation on short- and long-chain PA–modified and original silicone rubber surfaces on the esophageal side of the voice prosthesis.
The planimetrical biofilm scores of the surfaces of all 3 short-chain PA–treated voice prostheses indicated more biofouling on the treated surfaces than on the untreated surfaces of the same prostheses. For the long-chain PA–treated prostheses, the planimetrical biofilm scores, as well as the numbers of colony-forming units per cm−2 for bacteria and yeasts, indicated less biofouling on the treated side than on the control side for 9 of the 13 prostheses that could be analyzed (2 were lost to analysis). Identical fungal strains, mainly Candida sp, were isolated from biofilms on each side of the esophageal flange.
Chemisorption of long-chain PAs by the silicone rubber used for voice prostheses reduces biofilm formation in vivo and therefore can be expected to prolong the life of these prostheses. Chemisorption of short-chain PAs by silicone rubber seems to have an adverse effect.
TRACHEOESOPHAGEAL puncture voice prostheses have become the most frequent method of postlaryngectomy voice rehabilitation.1 A major drawback of these prostheses, however, involves their colonization within several weeks by a thick biofilm of adhering yeast and bacterial strains, causing increased airflow resistance and impairing the valve function.2- 5 As a consequence, indwelling silicone rubber voice prostheses such as the Groningen button (Medin Medical Instruments and Supplies, Groningen, the Netherlands), Provox (Atos Medical AB, Hörby, Sweden), or Blom-Singer (InHealth Technologies, Carpinteria, Calif) indwelling voice prostheses have to be replaced on average every 3 to 4 months.6- 9 Analysis of the biofilms on prostheses removed from patients demonstrated that the colonizing yeast strains were often Candida albicans and Candida tropicalis.2,3,8,10 Bacterial strains identified were of oral origin, including Streptococcus mitis, Streptococcus sobrinus, and Streptococcus salivarius, or were commensals from skin, such as Staphylococcus epidermidis and other staphylococcal isolates.8
Although indwelling voice prostheses are easily replaced with an outpatient-clinic procedure, frequent changes are distressing for the patients and increase the cost of health care. Various attempts have been made to decrease biofilm formation on indwelling voice prostheses, with varying degrees of success. These attempts include daily intake of large amounts of dairy products such as Turkish yogurt or Kephir containing Streptococcus thermophilus and Lactobacillus bulgaricus strains, the use of a buccal bioadhesive slow-release tablet containing miconazole nitrate,11 and decontamination of the oropharyngeal cavity from yeast.12 Also, up to 0.5% to 9% (weight-to-weight ratio) rifampicin has been incorporated in the silicone rubber,13 but this affected the mechanical properties of the polymer.
Microbial adhesion to biomedical implants, including voice prostheses, is determined by the properties of the biomaterial's surface. An alternative method to prevent or decrease biofilm formation might be to modify the silicone rubber surface. Recently, we have shown that the hydrophobicity of this surface influenced microbial adhesionin vitro14,15 and biofilm formation on voice prostheses in vivo, even after 4 weeks' use.16,17 Further, chemisorption of PAs substantially reduced microbial adhesion to silicone rubber in vitro.15 Moreover, microorganisms were easily detached from silicone rubber surfaces with chemisorbed, long-chain PAs, making this modification promising for the preparation of low-fouling, silicone rubber voice prostheses.
It is the aim of this article to compare the biofilm formation on silicone rubber Groningen button voice prostheses in vivo, with and without chemisorbed PAs.
"Ultra-Low Resistance" Groningen button voice prostheses (Medin Medical Instruments and Supplies) were used in this study. The silicone rubber was modified in a 2-step process, as schematically shown in Figure 1. One half of the esophageal flange of each voice prosthesis was oxidized in an argon-plasma treatment.17 In the second step, perfluoro-alkyltrichlorosilanes (Fluka Chemie AG, Buchs, Switzerland) were chemisorbed onto the oxidized surfaces by immersion for 10 minutes in 0.5% perfluoro-alkyltrichlorosilanes dissolved in perfluoroheptane, leaving one half of the esophageal flange untreated. Perfluoroheptane was chosen because it does not swell the silicone rubber. We used 2 different silanes with short (1 fluorocarbon unit) and long (8 fluorocarbon units) fluorocarbon chains, respectively. Silane-treated surfaces were washed with perfluoroheptane and absolute ethanol. Mean (SD) advancing water contact angles on the sides of the silicone rubber prostheses treated with chemisorbed short and long PA chains were 125 (5) and 140 (5) degrees, respectively,15 which is higher than on the untreated sides (115 (±3) degrees), signifying a higher degree of hydrophobicity of the treated surface. Chemisorption of PA chains did not adversely affect the biocompatibility of the silicone rubber, according to an approved agar diffusion test (BSC105/2; Bioscan BV, Bilthoven, the Netherlands). The airflow resistance of the partially modified Groningen button voice prostheses was tested as described previously18 and was identical to the airflow resistance of the original prostheses.
Following informed consent, each of 18 patients with laryngectomies and long-term experience with the use of a voice prosthesis was given a partially modified (by chemisorption of either short- or long-chain PAs) Groningen button voice prosthesis. All patients were randomly included in the study with no selection criteria other than that they required replacement of a prosthesis for medical reasons of leakage or increased airflow resistance. The tested voice prostheses were removed at the following regular outpatient clinic visit. After removal of the prosthesis from the tracheo-esophageal shunt, biofilm formation on the modified and unmodified sides was compared by light microscopy, and a planimetric biofilm score was calculated as the percentage of the surface (of the esophageal flange) colonized by microorganisms. Subsequently, culture samples were taken and the prostheses prepared for scanning electron microscopy. Microbial compositions of the biofilms on both sides of the valve were compared by plating on brain-heart infusion and blood agar plates (OXOID, Basingstoke, England) and by culturing at 37°C under aerobic conditions.8 Also, the number of colony-forming units of bacteria and yeasts per unit area (cm2) was determined for each side of the prostheses as a second, quantitative measure for biofilm formation.
After evaluation of the first 3 prostheses with the short PAs (1 fluorocarbon unit), it became obvious that the surface-modified part of the esophageal flange attracted approximately twice the amount of biofilm as the untreated part. Therefore, it was decided to stop the experiment with the short PAs.
Figure 2 shows a scanning electron micrograph of the esophageal flange of a partially long-chain PA surface-modified Groningen button prosthesis after 4 weeks of use. On the modified side, long PA chains (8 fluorocarbon units) had been chemisorbed to the silicone rubber. Fewer microcolonies formed on the modified side than on the original silicone rubber side of the prosthesis.
Table 1 summarizes the planimetric biofilm scores and the numbers of CFUs per unit area cm2 on each side of the voice prostheses after chemisorption of the long-chain PAs. One voice prosthesis was lost to microbiological analysis because the patient did not show up for routine follow-up but had the experimental voice prosthesis changed outside of office hours, 6 months after it had been inserted. Another prosthesis is still in place in a patient who chose to keep it; this prosthesis has now been situated for more than a year. Chemisorption of PAs with 8 fluorocarbon units yielded a reduction in planimetric biofilm scores and an accompanying reduction in CFUs/cm2 in 9 of the 13 patients. In 2 patients, an increase in planimetric biofilm score occurred.
No shift in yeast composition of the biofilms occurred as a result of the surface modification by either short- or long-chain PAs, with C albicans being a commonly found yeast strain. However, generally fewer different bacterial strains were isolated from the surface-modified sides than from the original silicone rubber sides, although no new strains were found on the modified sides. Staphylococci were the most commonly isolated bacterial strains, together with S mitis, Micrococcus luteus, and Rothia dentrocariosa.
In this study we evaluated the potential of PA chemisorption on silicone rubber voice prostheses as a means to reduce biofilm formation and thereby prolong the prostheses' life. As previously described,16 the use of partially surface-modified prostheses (each with an untreated control part) precludes the need to control all external factors that might influence a comparison of biofilm formation on both sides (eg, humidity, air temperature, nutritional factors, and other variations in the patient's lifestyle). As a consequence, the method allows conclusions to be drawn after a limited number of clinical trials. Moreover, the "split-button" method used here may be the only method to evaluate biofilm formation on surface-modified voice prostheses in vivo. However, one cannot draw conclusions about the extension of prosthesis life because these prostheses were only partly modified.
The results of this study indicate that chemisorption of long-chain PAs on silicone rubber yields reduced biofilm formation with possibly increased effects when longer fluorocarbon chains are used. The effects of the surface modification are obvious from the number of CFUs isolated per unit area cm2 and from the planimetric biofilm scores. The planimetric biofilm score does not differentiate between thick and thin biofilms once the score has reached 100% and a single film of adhering microorganisms has formed. Therefore, the planimetric biofilm score is a measure of the extension of biofilm over the esophageal side of the prosthesis and may have great relevance in predicting the occurrence of valve failure during use.
The reduction of biofilm formation on the long-chain PA–modified silicone rubber is a result of a variety of factors. First, the increased hydrophobicity of surface-modified silicone rubber contributes to the reduction in biofilm formation, as also demonstrated for dental plaque formation in the oral cavity.19 The reduction in dental plaque formation through increased substratum hydrophobicity involves bacteria more than yeasts. This is in line with the present results, demonstrating reductions in bacterial colonization of treated voice prostheses by several orders of magnitude. In comparison, reductions in colonization by yeasts were minor (Table 1), although in 3 of the 13 patients, no yeasts were isolated from the modified sides. In vitro studies have indicated, however, that only mixed colonization of silicone rubber by bacteria and yeasts yields the typical detrimental ingrowth of microcolonies causing dysfunction of prostheses.20,21 For the present application, increased hydrophobicity due to fluoridation of the silicone rubber alone is not sufficient protection against bacterial colonization; chemisorption of short-chain PAs does not show a favorable effect on biofilm formation. We have proposed that chemisorbed PA chains on silicone rubber form a dendritic wedge structure15 with a high degree of freedom of the chemisorbed chains. The favorable effects of chemisorbed long-chain PAs compared with short-chain PAs can be attributed to a higher mobility of the chemisorbed polymer chains, in a sense repelling the biofilm. Hence it is anticipated that the use of even longer fluorocarbon chains may show even greater beneficial effects.
This study demonstrates that biofilm formation on silicone rubber Groningen button voice prostheses over an evaluation period of approximately 2 to 8 weeks can be reduced by chemisorption of long (8 fluorocarbon units) PA polymer chains owing to the high hydrophobicity and mobility of these chemisorbed chains. Whether the average life of indwelling voice prostheses can also be prolonged by the modification studied here remains to be determined.
Accepted for publication July 23, 1999.
This project was funded in part by a grant from the European Community, Eureka project EU72310.
We thank Medin Instruments and Supplies (Groningen, the Netherlands) for providing the Groningen button voice prostheses used in this study.
Reprints: Henk J. Busscher, MSC, PhD, Department of Biomedical Engineering, University of Groningen, Bloemsingel 10, 9712 KZ Groningen, the Netherlands.