Study Design
The TLC trial randomized 780 children aged 12–33 months with blood lead levels of 20–44 µg/dL to succimer or placebo across four clinical sites. This section examines the trial’s design, its documented limitations, and the blood lead data it produced.
Enrollment by Race & Clinical Center
| Race | Baltimore | Newark | Philadelphia | Cincinnati | Total | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| White | 12 | 5.6% | 4 | 1.9% | 6 | 3.6% | 67 | 34.5% | 89 | 11.4% |
| Black | 186 | 87.3% | 146 | 70.2% | 143 | 86.7% | 120 | 61.9% | 595 | 76.3% |
| Hispanic† | 1 | 0.5% | 42 | 20.2% | 4 | 2.4% | 1 | 0.5% | 48 | 6.2% |
| Asian | 0 | 0.0% | 1 | 0.5% | 1 | 0.5% | 0 | 0.0% | 2 | 0.3% |
| Native American | 1 | 0.5% | 1 | 0.5% | 0 | 0.0% | 1 | 0.5% | 3 | 0.4% |
| Other | 5 | 2.3% | 14 | 6.7% | 7 | 4.2% | 4 | 2.1% | 30 | 3.8% |
| Unknown | 8 | 3.8% | 0 | 0.0% | 4 | 2.4% | 1 | 0.5% | 13 | 1.7% |
| Total | 213 | 208 | 165 | 194 | 780 | |||||
† Only the Newark clinical center had Spanish language staff and consent documents.
Enrollment began September 7, 1994 and was completed January 22, 1997.
Source: NIEHS TLC Trial Website (archived 1999)
Limitations
Seven methodological limitations that complicate interpretation of the trial’s null findings — including uncontrolled environmental exposure, unreported iron status, withheld adherence data, and untested site interactions.
Protocol, Annotated
Side-by-side readings of Protocol v9 (August 1994, pre-randomization) and Protocol v10 (November 1997, post-treatment), with scholarly commentary drawn from primary sources. Starting with blinding & placebo comparability.
Blood Lead Kinetics
Blood lead levels in TLC participants did not follow the trajectory the trial was designed to produce. Instead of a sustained separation between treatment arms, the data show a brief drop in the succimer group followed by rapid rebound and convergence with placebo — against a backdrop of rising blood lead levels in both groups during the first months after randomization.
Sources: Treatment-group BLL values at baseline, day 7, day 42, 6 months, 36 months, and 60 months from Chen et al. 2006 (Environ Health Perspect 114:579–583), Figure 1 caption. Intermediate treatment-phase values interpolated from Safety & Efficacy (Pediatr Res 2000;48:593–599) Figure 1 curve descriptions. Peak BLL (30.6 µg/dL) from Chen et al. 2005 (Environ Health Perspect 113:597–601), Table 1. CDC reference value (3.5 µg/dL) current as of 2021.
The Post-Randomization Spike
Chen et al. (2005) reported that the mean peak blood lead level across all 780 children was 30.6 µg/dL — a 4.4 µg/dL increase from the baseline mean of 26.2 µg/dL. This peak occurred at a mean age of 2.2 years, approximately 2.4 months after randomization. Both treatment arms experienced this spike.
Children got more lead-poisoned during the trial, not less. The one-time home cleaning provided before randomization was designed to suppress lead dust for approximately six months. It was insufficient. Ettinger et al. (2002) documented that many homes still had dust lead levels above federal standards after cleaning, and Farfel et al. (2000) showed dust lead loadings in Baltimore TLC homes rebounded to near-baseline within 24 months.
Treatment Phase: Separation and Convergence
Succimer produced an immediate and dramatic drop in blood lead — approximately 11 µg/dL within the first week of treatment. But blood lead levels rebounded within days of stopping each course, presumably from lead stored in bone re-entering circulation. The sawtooth pattern repeated with second and third courses.
The mean blood lead level of the succimer group was 4.5 µg/dL (95% CI, 3.7–5.3) lower than the placebo group over the first 6 months, and 2.7 µg/dL (95% CI, 1.9–3.5) lower over 12 months (Rogan et al., 2001). By 36 months, the groups were indistinguishable. The sustained 10 µg/dL separation the trial was powered to detect never materialized.
Meanwhile, the placebo group’s blood lead levels declined on their own — likely from the initial home cleaning, growth-related serum dilution, and natural deposition of lead into bone. Both arms converged around 12 µg/dL at age 5 and 8 µg/dL at age 7 — still more than double the current CDC reference value of 3.5 µg/dL.
Concurrent vs. Peak Blood Lead
The TLC trial was designed around the then-prevailing assumption that peak childhood blood lead (usually occurring around age 2) was the strongest predictor of later cognitive deficits. The treatment protocol aimed to lower that peak.
But TLC’s own secondary analyses contradicted this assumption. Chen et al. (2005) found that concurrent blood lead — measured at the same time as IQ — was the strongest predictor of cognitive deficits, not peak blood lead from age 2. The cross-sectional association between BLL and IQ strengthened as children aged, while the relationship with baseline BLL attenuated.
This finding was corroborated by the Lanphear et al. (2005) pooled analysis, which also found concurrent BLL to be the strongest correlate of IQ loss.
The implication: if concurrent blood lead matters most, then a treatment protocol that fails to produce sustained BLL reduction is testing the wrong thing. TLC lowered blood lead temporarily and measured IQ years later, when the groups were identical. The null result may say more about the protocol’s inability to maintain separation than about chelation’s inability to help.
Source documents referenced on this page are available in the TLC Reference Library.