What Does a Scientist “Gain,” When His Signal Is Only Noise
When the silicone litigation erupted in the early 1990s, Leoncio Garrido was a research professor at Harvard. In 1995, he was promoted from Assistant to Associate Professor of Radiology, and the Associate Director of NMR Core, at the Harvard Medical School. Along with Bettina Pfleiderer, Garrido published a series of articles on the use of silicon 29 nuclear magnetic resonance (NMR) spectroscopy, in which he claimed to detect and quantify silicon that migrated from the silicone in gel implants to the blood, livers, and brains of implanted women[1].
Plaintiffs touted Garrido’s work on NMR silicone as their “Harvard” study, to offset the prestige that the Harvard Nurses epidemiologic study[2] had in diminishing the plaintiffs’ claims that silicone caused autoimmune disease. Even though Garrido’s work was soundly criticized in the scientific literature[3], Garrido’s apparent independence of the litigation industry, his Harvard affiliation, and the difficulty in understanding the technical details of NMR spectroscopic work, combined to enhance the credibility of the plaintiffs’ claims.
Professor Peter Macdonald, who had consulted with defense counsel, was quite skeptical of Garrido’s work on silicone. In sum, Macdonald’s analysis showed that Garrido’s conclusions were not supported by the NMR spectra presented in Garrido’s papers. The spectra shown had signal-to-noise ratios too low to allow a determination of putative silicon biodegradation products (let alone to quantify such products), in either in vivo or ex vivo analyses. The existence of Garrido’s papers in peer-reviewed journals, however, allowed credulous scientists and members of the media to press unsupported theories about degradation of silicone into supposedly bioreactive silica.
A Milli-Mole Spills the Beans on the Silicone NMR Data
As the silicone litigation plodded on, a confidential informant dropped the dime on Garrido. The informant was a Harvard graduate student, who was quite concerned about the repercussions of pointing the finger at the senior scientist in charge of his laboratory work. Fortunately, and honorably, this young scientist more concerned yet that Garrido was manipulating the NMR spectra to create his experimental results. Over the course of 1997, the informant, who was dubbed “Mini-Mole,” reported serious questions about the validity of the silicon NMR spectra reported by Garrido and colleagues, who had created the appearance of a signal by turning up the gain to enhance the signal/noise ratio. Milli-mole also confirmed Macdonald’s suspicions that Garrido had created noise artifacts (either intentionally or carelessly) that could be misrepresented to be silicon-containing materials with silicon 29 NMR spectra.
In late winter 1997, “Mini-Mole” reported that Harvard had empanelled an internal review board to investigate Garrido’s work on silicon detection in blood of women with silicone gel breast implants. The board involved an associate dean of the medical school, along with an independent reviewer, knowledgeable about NMR. Mini-Mole was relieved that he would not be put into the position of becoming a whistle blower, and he believed that once the board understood the issues, Garrido’s deviation from the scientific standard of care would become clear. Apparently, concern at Harvard was reaching a crescendo, as Garrido was about to present yet another abstract, on brain silicon levels, at an upcoming meeting of the International Society of Magnetic Resonance in Medicine, in Vancouver, BC. Milli-Mole reported that one of the co-authors strongly disagreed with Garrido’s interpretation of the data, but was anxious about withdrawing from the publication.
Science Means Never Having to Say You’re Sorry
By 1997, Judge Pointer had appointed a panel of neutral expert witnesses, but the process had become mired in procedural diversions. Bristol-Myers Squibb sought and obtained a commission in state court (New Jersey) cases for a Massachusetts’ subpoena for Garrido’s underlying data late in1997. Before BMS or the other defendants could act on this subpoena, however, Garrido published a rather weak, non-apologetic corrigendum to one of his papers[4].
Although Garrido’s “Erratum” concealed more than it disclosed, the publication of the erratum triggered an avalanche of critical scrutiny. One of the members of the editorial board of Magnetic Resonance in Medicine undertook a critical review of Garrido’s papers, as a result of the erratum and its fallout. This scientist concluded that:
“From my viewpoint as an analytical spectroscopist, the result of this exercise was disturbing and disappointing. In my judgement as a referee, none of the Garrido group’s papers (1–6) should have been published in their current form.”
William E. Hull, “A Critical Review of MR Studies Concerning Silicone Breast Implants,” 42 Magnetic Resonance in Medicine 984, 984 (1999).
Another scientist, Professor Christopher T.G. Knight, of the University of Illinois at Urbana-Champaign, commented in a letter in response to the Garrido erratum:
“A series of papers has appeared in this Journal from research groups at Harvard Medical School and Massachusetts General Hospital. These papers describe magnetic resonance studies that purport to show significant concentrations of silicone and chemically related species in the blood and internal organs of silicone breast implant recipients. One paper in particular details 29Si NMR spectroscopic results of experiments conducted on the blood of volunteers with and without implants. In the spectrum of the implant recipients’ blood there appear to be several broad signals, whereas no signals are apparent in the spectrum of the blood of a volunteer with no implant. On these grounds, the authors claim that silicone and its degradation products occur in significant quantities in the blood of some implant recipients. Although this conclusion has been challenged, it has been widely quoted.
******
The erratum, in my opinion, deserves considerably more visibility, because it in effect greatly reduces the strength of the authors’ original claims. Indeed, it appears to be tantamount to a retraction of these.”
Christopher T.G. Knight, “Migration and Chemical Modification of Silicone in Women With Breast Prostheses,” 42 Magnetic Resonance in Med. 42:979 (1999) (internal citations omitted). Professor Knight went on to critique the original Garrido work, and the unsigned, unattributed erratum as failing to show a difference between the spectra developed from blood of women with and without silicone implants. Garrido’s erratum suggested that his “error” was simply showing a spectrum with the wrong scale, but Professor Knight showed rather conclusively that other manipulations had taken place to alter the spectrum. Id.
In a brief response[5], Garrido and co-authors acknowledged that their silicon quantification was invalid, but still maintained that they had qualitatively determined the presence of silicon entities. Despite Garrido’s response, the scientific community soon became incredulous about his silicone NMR work.
Garrido’s fall-back claim that he had detected unquantified levels of silicon using Si29 NMR was definitively refuted, in short order[6]. Ultimately, Peter Macdonald’s critique of Garrido was vindicated, and Garrido’s work became yet another weight that helped sink the plaintiffs’ case. Garrido last published on silicone in 1999, and left Harvard soon thereafter, to become the Director of the Instituto de Ciencia y Tecnología de Polímeros, in Madrid, Spain. He is now a scientific investigator at the Institute’s Physical Chemistry of Polymers Department. The Institute’s website lists Garrido as Dr. Leoncio Garrido Fernández. Garrido’s silicone publications were never retracted, and Harvard never publicly explained Garrido’s departure.
[1] See, e.g., Bettina Pfleiderer & Leoncio Garrido, “Migration and accumulation of silicone in the liver of women with silicone gel-filled breast implants,” 33 Magnetic Resonance in Med. 8 (1995); Leoncio Garrido, Bettina Pfleiderer, B.G. Jenkins, Carol A. Hulka, D.B. Kopans, “Migration and chemical modification of silicone in women with breast prostheses,” 31 Magnetic Resonance in Med. 328 (1994). Dr. Carol Hulka is the daughter of Dr. Barbara Hulka, who later served as a neutral expert witness, appointed by Judge Pointer in MDL 926.
[2] Jorge Sanchez-Guerrero, Graham A. Colditz, Elizabeth W. Karlson, David J. Hunter, Frank E. Speizer, Matthew H. Liang, “Silicone Breast Implants and the Risk of Connective-Tissue Diseases and Symptoms,” 332 New Engl. J. Med . 1666 (1995).
[3] See R.B. Taylor, J.J. Kennan, “29Si NMR and blood silicon levels in silicone gel breast implant recipients,” 36 Magnetic Resonance in Med. 498 (1996); Peter Macdonald, N. Plavac, W. Peters, Stanley Lugowski, D. Smith, “Failure of 29Si NMR to detect increased blood silicon levels in silicone gel breast implant recipients,” 67 Analytical Chem. 3799 (1995).
[4] Leoncio Garrido, Bettina Pfleiderer, G. Jenkins, Carol A. Hulka, Daniel B. Kopans, “Erratum,” 40 Magnetic Resonance in Med. 689 (1998).
[5] Leoncio Garrido, Bettina Pfleiderer, G. Jenkins, Carol A. Hulka, Daniel B. Kopans, “Response,” 40 Magnetic Resonance in Med. 995 (1998).
[6] See Darlene J. Semchyschyn & Peter M. Macdonald, “Limits of Detection of Polydimethylsiloxane in 29Si NMR Spectroscopy,” 43 Magnetic Resonance in Med. 607 (2000) (Garrido’s erratum acknowledges that his group’s spectra contain no quantifiable silicon resonances, but their 29Si spectra fail to show evidence of silicone or breakdown products); Christopher T. G. Knight & Stephen D. Kinrade, “Silicon-29 Nuclear Magnetic Resonance Spectroscopy Detection Limits,” 71 Anal. Chem. 265 (1999).
