Late Seromas in Natrelle 410 Form-Stable Silicone Breast Implants
The authors discuss the rate of late seroma and they state that “…as the rate observed in this study is similar to that reported for breast implants in general, it is unlikely that the Natrelle 410 implant itself confers an incremental risk of late seroma relative to other implants.” The authors give no reference for this claim and ignored Spear’s article that concluded “Biocell textured implants were more likely to be associated with late seromas than were smooth shell implants.”3 Textured implants have been involved in 55 of 60 cases in the review by Park et al.,4 six of six cases in the series reported by Pinchuk and Tymofii,5 eight of eight cases in the series by Mazzocchi et al.,6 and 27 of 28 cases in the series by Spear et al.3 Although the causal relationship between textured implants and late seroma has not yet been elucidated and despite the limitations of these studies (discussed below), this association is very strong. It does not seem appropriate to implicate that the Natrelle 410 breast implant, which is textured, is not as “incrementally risky” as nontextured implants.
I (E.J.H.F.) have been in practice for 34 years, and colleagues of my generation rarely if ever saw a late seroma with a smooth implant. The problem is now highlighted in international conferences where whole panels are being dedicated to the discussion of late seromas (e.g., the Latin American Society of Plastic Surgeons, in Panama City, Panama, June 1 through 4, 2010), and is a recent phenomenon that began when textured implants were introduced and almost universally adopted.
All case series to date on late seromas have been subject to lead-time bias with nonuniform exposure times between subjects that were not adjusted for in analysis (i.e., unadjusted estimates of late seroma in subjects with shorter follow-up times causes underestimation of the true rate of late seroma). In this article, 16 of 26 late seromas were reported under the mean follow-up time, with a skew toward seromas under 3 years. The true rate of late seromas is not well-estimated given this limitation and uncontrolled exposure time. The association between implant age (it was not clear whether implant age was the same as time after implantation) and late seroma supports this inherent bias and underestimated rate.
Lastly, the authors did not include any time-based outcomes, but reported a Cox proportional hazards regression model. This makes us wonder what multivariate models were being used to calculate adjusted risk ratios. The model from the “backward elimination technique” was not presented, and postregression analysis, including discussion of possible interaction and confounding, was not reported.
We feel that the risk of late seroma is likely underestimated in the literature. Controlling for patient exposure time should be required in postmarketing surveillance database analysis to avoid issues such as lead-time bias.