Release of Low Molecular Weight Silicones
                                 and Platinum from Silicone
                         Ernest D. Lykissa, Subbarao V. Kala, Jennifer B. Hurley, and Russell M. Lebovitz*
                            Department of Pathology, Baylor College of Medicine, Houston, Texas 77030

                                    Received for review July 7, 1997. Accepted September 30, 1997.
                          .Anal. Chem., 69 (23), 4912 -4916, 1997. ac970710w S0003-2700(97)00710-5
                                                Copyright 1997 American Chemical Society


We have conducted a series of studies addressing the chemical composition of silicone gels from breast implants as well as the diffusion of low molecular weight silicones (LM-silicones) and heavy metals from intact implants into various surrounding media, namely, lipid-rich medium (soy oil), aqueous tissue culture medium (modified Dulbecco's medium, DMEM), or an emulsion consisting of DMEM plus 10% soy oil. LM-silicones in both implants and surrounding media were detected and quantitated using gas chromatography (GC) coupled with atomic emission (GC-AED) as well as mass spectrometric (GC/MS) detectors, which can detect silicones in the nanogram range.

Platinum, a catalyst used in the preparation of silicone gels, was detected and quantitated using inductive argon-coupled plasma/mass spectrometry (ICP-MS), which can detect platinum in the parts per trillion range. Our results indicate that GC-detectable low molecular weight silicones contribute approximately 1-2% to the total gel mass and consist predominantly of cyclic and linear poly(dimethylsiloxanes) ranging from 3 to 20 siloxane [(CH3)2-Si-O] units (molecular weight 200-1500). Platinum can be detected in implant gels at levels of ~700 [Image]g/kg by ICP-MS.  The major component of implant gels appears to be high molecular weight silicone polymers (HM-silicones) too large to be detected by GC. However, these HM-silicones can be converted almost quantitatively (80% by mass) to LM-silicones by heating implant gels at 150-180 [Image]C for several hours.

We also studied the rates at which LM-silicones and platinum leak through the intact implant outer shell into the surrounding media under a variety of conditions. Leakage of silicones was greatest when the surrounding medium was lipid-rich, and up to 10 mg/day LM-silicones was observed to diffuse into a lipid-rich medium per 250 g of implant at 37 [Image]C. This rate of leakage was maintained over a 7-day experimental period. Similarly, platinum was also observed to leak through intact implants into lipid-containing media at rates of approximately 20-25 [Image]g/day/250 g of implant at 37 [Image]C. The rates at which both LM-silicones and platinum have been observed to leak from intact implants could lead to significant accumulation within lipid-rich tissues and should be investigated more fully in vivo.

Silicones are polymeric organic siloxanes consisting of a backbone of alternating silicon-oxygen [Si-O] units with organic side chains attached to each silicon atom.1,2 Silicones may be polymerized through a variety of chemical reactions to form chains of varying lengths. As chain length is increased, the polymer becomes increasingly viscous and hydrophobic. Cross-linking of silicone polymers leads to formation of gels or rubber-like solids. The great range of viscosity exhibited by silicone oils and the elasticity of silicone rubbers have made it possible to fabricate silicone-based materials that mimic the consistency of tissues ranging from bone to mammary tissue.1,2  Poly(dimethylsiloxane) (PDMS) is the most common siloxane polymer used in medical products, including breast implants. This material was chosen for medical applications because it was believed to be biologically inert.1,2

While a great deal of evidence supports the claim that high molecular weight PDMS polymers are relatively inert in biological systems, the same studies strongly suggest that certain LM-silicones, particularly the cyclic silicones, may have potent biological activities that mimic, for example, estrogens2,3 or CNS-active drugs.2,4 We have recently developed
novel and highly sensitive methods for detecting, quantitating, and characterizing     LM-silicones in biological tissues using GC-AED and GC/MS.5 Using GC-AED and GC/MS in the present study, we have detected significant levels of LM-siloxanes within the breast implant PDMS gels and have demonstrated that LM-siloxanes appear to diffuse readily through intact implant elastomer shells, particularly into lipid-rich media.

Since platinum, which is used commonly as a catalyst for cross-linking of silicone polymers to form gels and silicone rubbers,1 also exhibits significant toxicity6,7 in vivo and in vitro, we have also addressed the overall levels of platinum in implant gels as well as the ability of platinum-containing compounds to leak from intact implants into surrounding media at physiologic temperatures using ICP-MS, which offers sensitivity in the parts per trillion (ng/kg) range.8

Materials and Methods

Reagents. Ethyl acetate, chromatography grade, OmniSolv, was purchased from EM Science; activated coconut charcoal, 6-14 mesh was purchased from Fisher Scientific; soybean oil contains 0.64 g of polyunsaturated fat, 0.21g of monounsaturated fat, and 0.15 g of saturated fat per gram of oil. Modified Dulbecco's medium (DMEM) was purchased from Sigma Chemical (St. Louis, MO).

Measurement of Silicones. Gas chromatography/mass spectroscopy (GC/MS) analysis of silicones was carried out as described previously5 using a Hewlett Packard Model 5972 gas chromatography/mass spectrometry system Hewlett Packard, Palo Alto, CA) equipped with a Windows 95 HP  workstation capable of performing a full spectral search using the Wiley Spectral Library. GC-AED analysis of silicones was performed5 using a Hewlett Packard 5890/5921A gas chromatography/atomic emission detection system equipped with a Pascal workstation. Extra low bleed DB-XLB (J&W Scientific, Folsom, CA) capillary columns were used for the chromatographic separation of silicones in order to minimize the column bleed of hexamethylcyclotrisiloxane. Breast implant gels from explanted implants were dissolved in ethyl acetate (1g/10 mL) and analyzed for the presence of low molecular weight silicones using GC-AED and GC/MS. Quantitation of low molecular cyclic siloxanes was carried out with external standard calibration curves obtained using hexamethylcyclotrisiloxane (D3), octamethylcyclotetrasiloxane (D4), decamethylcylopentasiloxane (D5), and dodecamethylcyclohexasiloxane (D6)  obtained from Ohio Valley Speciality Chemical, Marietta, OH. Ethyl acetate blanks were run between samples to avoid possible carry-over during the analysis, and the blank values were subtracted from the sample values during the data analysis.

Leakage Studies with Intact Breast Implants. Leakage of low molecular weight poly(dimethylcyclosiloxanes) and platinum was determined using explanted  breast implants. In all cases, the implants were carefully checked to exclude any damage to the outer envelope and were wiped until no LM-silicones could be detected on the outer shell surface by GC/MS. Intact breast implants were weighed and placed individually in tightly sealed glass containers containing lipid-rich medium (soy oil), aqueous tissue culture medium (DMEM), or an emulsion consisting of DMEM plus 10% soy oil. After 24 h at 37 oC, the medium was removed and extracted with (2:1) ethyl acetate/methanol for GC analysis of LM-silicones. To evaluate the release of LM-silicones from intact implants into air at 37 [Image]C, activated charcoal was used to adsorb siloxanes released into the air-filled head space above the implants, followed by ethyl acetate extraction of the charcoal.

The implants used in this study weighed between 240 and 260 g. At least three different implants were used for each determination and all measurement were performed in duplicate or triplicate. Twenty-four-hour release of LM-silicones was measured daily over a 7-day period to assess the linearity of the release. Corresponding extracts of media without implants were used as blanks.

Platinum Detection. Platinum levels were evaluated using a Hewlett Packard Model 4500 ICP-MS system (Hewlett Packard) equipped with a crossflow-type nebulizer.8 Oxygen was added as a carbon scavenger into the sample gas
at 8% of the total sample gas flow. A calibration curve was made using 50:50 1-propanol/water as solvent. Pb was added on-line in 50:50 1-propanol/water as an internal standard at the ppb level. The samples were diluted in 1-propanol and aspirated into the ICP-MS for the analysis of platinum; the values are expressed as micrograms of platinum/per kilogram (ppb).


LM-Silicone Components in Implant Gels. Breast implant gels dissolve readily in 10 volumes of ethyl acetate, and analysis of LM-silicones contained in unfractionated implant gel by GC-AED demonstrates that approximately 25-30 different LM-silicone species can be resolved . Further  analysis of this material by GC/MS and comparison with the Wiley Mass-spectral Library indicates that the five earliest-eluting chromatographic peaks correspond to cyclic PDMS polymers with 3-7 PDMS subunits. The sixth chromatographic peak  corresponds to a linear PDMS polymer with 7 PDMS subunits. Later-eluting peaks cannot be identified unambiguously by comparison with the Wiley Library, since our MS instrument only identifies fragments of m/z 600 or  less. However, linear and cyclic silicones can be distinguished by mass
spectroscopy based upon the characteristic mass spectra of these compounds (the m/z 221 is most abundant in the linear siloxane molecules as opposed to cyclics) (Lykissa, Kala, Hurley, and Lebovitz, unpublished). The chromatographic peaks corresponding to linear LM-silicones are marked with an asterisk in Figure 1. It is clear from Figure 1 that both linear and cyclic LM-silicones are present, and they elute as an alternating series  (cyclic followed by corresponding linear of the same chain length). Although we interpret Figure 1 as demonstrating LM-silicones up to D20 and L20, only those peaks actually identified by direct matches with the Wiley Library are labeled. No other significant chromatographic peaks were visible by either GC-AED or GC/MS, suggesting that LM-silicones represent the only low molecular weight, volatile organic molecules within the implant gel.

Figure 1 Analysis of low molecular weight siloxanes (LM Silicones) in silicone implant gels using GC-AED. Implant gels were solubilized in 10 volumes of ethyl acetate, and 100 [Image]g of total implant gel was fractionated by GC. Thirty-two distinct silicon-containing peaks were detected by AED at 251.6 nm. Linear  LM-silicones are marked with an asterisk, while cyclics are [Image] unmarked. Peaks 1-6 have been identified as D3-D7 and L7 (hexadecamethylheptasiloxane), respectively, by comparison with the Wiley Mass Spectral Library using GC/MS. The remaining peaks could not be identified unambiguously since our mass spectrometer could only detect ion fragments with m/z < 600 amu. However,
we can distinguish the linear and cyclic siloxanes based upon the  pattern of ion fragmentation (see text).

Quantitation of the LM-silicone peaks and comparison with the total mass of gel used in our analysis indicates that LM-silicones account for ~1% of the total implant gel mass. The implants available to us for these studies  have  been explanted after 2-5 years of implantation, and the relative abundance of LM-silicones is likely to be significantly greater in virgin implants due to leakage of LM-silicones (see below). We have assumed, without direct evidence at this point, that the remaining 99% of silicone gel components consist largely of HM-silicones and cross-linked HM-silicones too large
to be separated using our capillary column GC instrumentation.

ICP-MS analysis of unfractionated breast implant gels indicates that platinum is present at a concentration of ~700 parts per billion (700 [Image]g/kg). Our ICP-MS analyses do not allow direct determination of whether platinum in implant gels is present as inorganic metal as opposed to organic or silicone-based complexes.

We attempted to purify the LM-silicones in implant gels by distillation.  Fractions of clear liquid could be produced directly from heated implant  gels at approximately 150-180 [Image]C using a constant vacuum of ~0.5 Torr to collect the heavy silicone vapors into chilled traps. Analysis of these fractions by GC/MS reveals that they consist predominantly (>98%) of D3-D8 with much smaller amounts of D9-D20 (Figure 2). The relative proportions of D3-D8 vary slightly with fractions distilled at different temperatures, but in all cases, D3-D5 is more abundant than D6-D8. After ~24 h of vacuum distillation, as much as 80% of the initial gel weight can be recovered as D3-D8. Based upon our earlier findings indicating that D3-D8 constitute on the order of 1% of total gel weight before distillation (Figure 1), the data in Figure 2 indicate that HM-silicones are converted almost quantitatively to LM-silicones after heating for extended periods at 150-180 [Image]C. Interestingly, the distillate contains ~40 parts per billion (40 [Image]g/kg) of platinum, suggesting that ~5% of the platinum in implant  gels resides in volatile organic or silicone-based complexes.

Figure 2 GC/MS analysis of distillation products from implant gels. Implant gels were heated directly to 150-180  Image]C and the vapor phase was collected and condensed under light vacuum  [Image] (~0.5 Torr). After several hours, up to 80% of the original gel mass could be recovered as distillate. This gas chromatogram represents 900 ng of implant distillate. Peaks 1-5 were identified as D3-D7 by comparison with the Wiley Mass Spectral  Library.

Diffusion of LM-Silicones through Intact Implant Outer Shells. Any potential toxicity of LM-silicones or platinum identified within implant gels is unlikely to be either biologically significant or medically relevant  unless these compounds are able to escape from the implant gel and outer shell into surrounding tissues. We have therefore attempted to measure the rates at
which both LM-silicones and platinum can leak from intact implants under a variety of different conditions. Figure 3 compares the total release of LM-silicones from intact implants into three different surrounding media  after incubation for 24 h at 37 [Image]C. The media used in this  experiment consist of a lipid-rich medium (soy oil), an aqueous cell culture medium  (DMEM), or an emulsion consisting of 90% DMEM/10% soy oil. LM-silicones were detected in all three media within 24 h, but the actual concentrations varied markedly among the different media. The highest concentration of  LM-silicones was found in the lipid-rich medium, which contained 40 [Image]g of total LM-silicones/g of implant gel after 24 h. In contrast, the aqueous DMEM contained less than 1 [Image]g of total LM-silicones/g of implant gel. These differences are due largely to the extremely low solubility of silicones in aqueous media, which rapidly reach saturation with
LM-silicones well before 24 h. The 90% DMEM/10% lipid emulsion contains ~4 [Image]g of total LM-silicones after 24 h. For each 250-g intact implant, our results indicate that as much as 10 mg of total LM-silicones can be released per
day into a lipid-rich environment such as might be found in human breast tissue. The observed 24-h release rates for  M-silicones are reduced by 75-85% after incubation at 4 [Image]C instead of 37 [Image]C (not shown), indicating
that the observed release of LM-silicones is probably dependent on diffusion rates through the implant shell.

Figure 3 Twenty-four-hour release of LM-silicones from intact implants. Implants were suspended in a closed jar filled with 500 mL of soy oil, 90% DMEM/10% soy oil, or DMEM for 24 h at 37 [Image]C, and the medium was subsequently extracted with ethyl  [Image] acetate and analyzed for the presence of LM-silicones by GC/MS. Values are expressed as micrograms of total LM-silicone released per gram of gel per day and represent the mean of six individual  determinations each from at least three different implants.

Figure 4A displays the GC/MS peak profile of LM-silicones released into lipid medium. D4-D7 appear to be released from implant gels more readily  than other LM-silicones, although larger species can also be detected. Figure 4B  ompares the individual 24-h release rates into lipid media for each of the LM-silicone components D4-D7. It is clear from Figures 3 and 4 that D4-D7 make up approximately 65% of the total LM-siloxanes released from intact implants. In addition, the release of D4-D7 into lipid-rich medium appears to be linear over a period of 7 days. No D3 was detected in oil,  probably because of its high volatility at 37 [Image]C. It is worth noting in this respect that D3 and D4 are released at relatively high concentrations from intact implants into the air at 37 [Image]C (data not shown) and that the observed concentrations of these molecules in lipid and aqueous media significantly underestimate their true release rates. For
example, D5 is detected in lipid-rich medium at levels greater than those observed for D4 (Figure 4A), even though D4 is present within the implant gel at higher concentrations than D5 (see Figure 1), and it is likely that this observation reflects the volatility of D4 at 37 [Image]C.

Figure 4 Release of individual LM-silicones from intact implants into lipid-rich medium. (A) GC/MS analysis of total
LM-silicones released into 500 mL of soy oil over a 24-h period at 37 [Image]C. Silicones were extracted from oil in ethyl [Image] acetate/methanol (see Materials and Methods). Peaks corresponding to linear LM-silicones are marked with an asterisk. Peaks  corresponding to decanoic and dodecanoic acids extracted from the oil are marked with a +. (B) Release of D4-D7 over a 7-day period. The oil medium was changed every 24 h for 7 days, and concentrations of D4-D7 were assayed by GC/MS.

We have also compared the release of platinum from intact implants into lipid-rich and aqueous media by ICP-MS, and the results are shown in Figure 5. 195Pt was detected at concentrations of 48 parts per billion (48 [Image]g/kg) after 24 h in lipid-rich medium and at 40 parts per billion (40 [Image]g/kg) in 10% lipid/90% DMEM. These findings correspond to total platinum release rates of 24 [Image]g/day in lipid-rich medium and 20 [Image]g/day in 90% DMEM/10% lipid. Platinum was undetectable in DMEM alone.

Figure 5 Release of platinum from intact implants into lipid-rich and aqueous media. Implants were suspended in a closed jar filled with 500 mL of soy oil, 90% DMEM/10% soy oil, or DMEM for 24 h at 37 [Image]C, and the medium was treated overnight with 0.1 N [Image] nitric acid and analyzed for the presence of platinum by ICP-MS. Values are expressed as microgram of total platinum released per kilogram of medium per day (ppb) and represent the mean of six
individual measurements each from at least three different implants.


Silicones have been used extensively for construction of a wide variety of   prosthetic medical devices, and the literature suggests that the HM-silicones and cross-linked HM-silicones are relatively stable and   inert in biological tissues.1,2 However, the LM-silicones, particularly the cyclic forms, exhibit a variety of potentially toxic biological activities
including relatively potent estrogen- and reserpine-like activities.2-4 Our current studies indicate that silicone breast implant gels contain as much as 2 g or greater of LM-silicones per 250-g implant. Since the implants used for our studies were explanted after several years of use, the levels of  LM-silicones in virgin implants may be even greater. We have also
demonstrated that the LM-silicones in implant gels consist predominantly of cyclic poly(dimethylsiloxanes) of molecular weight 200-1500, with lesser  amounts of linear poly(dimethylsiloxanes) of molecular weight 500-1500. The  M-siloxanes leak readily from intact implants at relatively constant rates of up to 10 mg per day per 250-g implant at 37 [Image]C; the rates of leakage are greatest when the implants are placed in a lipid-rich environment such as that found in breast tissue. With leakage rates comparable to those observed in our experiments, as much as 2 g of LM-siloxanes per year or greater could leak from a single intact implant and become distributed within lipid-rich tissues. Garrido et al.12 have noted
that the silicon content of adjacent breast tissue increases dramatically after implantation of silicone gel implants. Bergman et al.13 showed that filter paper in direct contact with intact implants increases in weight with increasing duration of contact. Our data not only support the findings of these earlier studies12-14 but also offer firm evidence as to the
identity, structure, and quantity of the particular LM-silicone being leaked. The predominant LM-silicones released from intact implants are D4-D8. Lower amounts of larger cyclic (D9-D20) and linear (L7-L20) silicones are also  released. Although D3 does not accumulate readily in lipid-rich media, it is clearly released from implants and represents the predominant LM-silicone species recovered from the vapor phase by activated charcoal.

Analysis of platinum levels in implant gels by ICP-MS indicates that each 250 g of implant contains a total of ~175 [Image]g of platinum at a concentration of 700 parts per billion (700 [Image]g/kg). Approximately 10-15% of this platinum (20-24 [Image]g) can be released into lipid-rich  medium after incubation at 37 [Image]C for 24 h. The ready release of platinum through the intact outer shell and into a lipid-rich environment suggests that at least a fraction of the implant platinum component may exist as an organoplatinum or silicone-platinum complex. The presence of  organoplatinum or silicone-platinum complexes in implant gels is also supported by data indicating that the distillation product of implant gels, which consists predominantly of cyclic LM-siloxanes (D3-D7), also contains 40 parts per billion of platinum (40 [Image]g/kg). If platinum is eventually shown to be complexed directly with LM-silicones, then the toxicity of these complexes in vivo and in vitro will need to be studied in greater detail.

One of the more interesting and surprising findings in this study is that although LM-silicones make up only ~1% by mass of the implant gel weight, as much as 80% of the total gel mass is released as LM-silicones after heating gels to 150-180 [Image]C under conditions of vacuum distillation. The resulting distillate consists predominantly of D3-D8, with very little
evidence of larger cyclic or linear siloxanes. The ready conversion of HM-silicones to cyclic LM-silicones at these moderate temperatures may be due to the presence of catalysts such as platinum within the implant and  raises the question of whether this conversion can also occur, albeit at a greatly reduced rate, after implantation at 37 [Image]C. Furthermore,
since the implant gel distillate contains predominantly D3-D7 as well as low concentrations of platinum, it closely  esembles the LM-silicone fraction released from intact implants at 37 [Image]C. We strongly suggest that  this distillate represents the most appropriate LM-silicone fraction available at present for testing of tissue distribution, biological activity, and
potential toxicity in experimental animal models and cell culture studies.

The documented leakage of LM-silicones and heavy metals from apparently intact implants raises troubling questions about the biological fate of these newly liberated compounds within tissues. Are LM-silicones transported to distant tissues on the basis of their lipid solubility? Are they metabolized and excreted, or do they accumulate in certain tissues selectively? What levels do they reach in various tissues with time, and can they exhibit toxicity or biological activity at these concentrations? What  is the fate of the platinum that is released? All of these important questions need to be addressed in the near future before we can fully and objectively assess whether or not silicone breast implants truly pose a
public health hazard.

Abbreviations: amu, atomic mass units; ppb, parts per billion; GC, gas chromatography; AED, atomic emission detection; MS, mass spectrometry; LM-silicones, low molecular weight silicones; HM-silicones, high molecular
weight silicones; m/z, mass/charge ratio.


The work described in this paper was supported by a grant from The Consumer Advocates for Product Safety (CAPS) Foundation. The authors also  acknowledge Joe Hedrick, Ken Helmreich, and Hewlett Packard for helping to arrange,
perform, and interpret the ICP-MS analyses.

* Corresponding author: (713)798-8161 (phone); (713)798-5838 (fax);

1. Potter, M.; Rose, N. R. Immunology of Silicones; Springer-Verlag: New  York, 1996.

2. LaVier, R. R.; Chandler, M. L.; Wendel, S. R. In Biochemistry of Silicon and Related Problems; Bendz, G., Lindquist, I., Eds.; Plenum Press: New York, 1978, pp 473-513.

3. Hayden, J. F.; Barlow, S. A. Toxicol. Appl. Pharmacol. 1972, 21, 68-79.

4. Carlisle, E. M. Science 1972, 178, 619-621.

5. Kala, S. V.; Lykissa, E. D.; Lebovitz, R. M. Anal. Chem. 1997, 69, 1267-1272.

6. Holbrook D. J., Jr.; Washington, M. E.; Leake, H. B.; Brubaker, P. E Environ. Health Perspect. 1975, 10, 95-101.

7. Carson, B. L.; Ellis, H. V., III; McCann, J. L. Toxicology and biological monitoring of metals in humans; Lewis Pubs Inc.: Chelsea, MI, 1986; pp 183-187.

8. Bakowska, E.; Potter, D. Hewlett Packard Application Note, 1995; pp 228-301.

9. Peters, W. J.; Smith, D.; Lugowski, S.; McHugh, A.; Keresteci A.; Baines, C. Ann. Plast. Surg. 1995, 34, 343-347.

10. Gitelman, H. J.; Alderman, F. R. J. Anal. Atom Spectrosc. 1990, 5, 587-589.

11. Evans, G. R. D.; Slezak, S.; Rieters, M.; Bercowy, G. M. Plast. Reconstr. Surg. 1994, 93, 1117-1122.

12. Garrido, L.; Pfeiderer, B.; Jenkins, B. G. Magn. Reson. Med. 1994, 31, 328-330.

13. Bergman, R.; Van der Ende, A. Br. J. Plast. Surg. 1979, 32, 31-34.

14. Yu, L.; LaTorre, G.; Marotta, J.; Batich, C.; Hardt, N. Plast. Reconstr. Surg. 1995, 97, 756-764.

Date: Tue, 23 Mar 1999 19:52:36 -0700
From: (Ilena Rose)
Organization: Humantics Foundation for Women