Power-laws and the use of pluripotent stem cell lines

Abstract

It is widely accepted that the (now reversed) Bush administration's decision to restrict federal funding for human embryonic stem cell (hESC) research to a few "eligible" hESC lines is responsible for the sustained preferential use of a small subset of hESC lines (principally the H1 and H9 lines) in basic and preclinical research. Yet, international hESC usage patterns, in both permissive and restrictive political environments, do not correlate with a specific type of stem cell policy. Here we conducted a descriptive analysis of hESC line usage and compared the ability of policy-driven processes and collaborative processes inherent to biomedical research to recapitulate global hESC usage patterns. We find that current global hESC usage can be modelled as a cumulative advantage process, independent of restrictive or permissive policy influence, suggesting a primarily innovation-driven (rather than policy-driven) mechanism underlying human pluripotent stem cell usage in preclinical research.

Figure 1. Analysis of the hESC co-citation network.

(A) Frequency distribution for hESC line use based on the evaluation of 2,338 studies reporting original research involving hESCs and published in peer-reviewed English language journals from 1998 to 2011. The inset highlights the 40 most used lines. Asterisks denote those hESC lines available and eligible for federal funding under the Bush administration from Aug. 9, 2001 to Mar. 9, 2009. Note that in most papers several hESC lines were used. (B) The largest connected network of the empirical hESC co-citation network. Peer reviewed studies involving experiments with identifiable hESC lines published from 1998 to 2011 are represented as boxes; hESC lines are represented as circles (see Fig. S2 for the entire network, including all disconnected components). The network is dominated by few lines (H1, H7, H9, HES-2, HES-3, BG01), which were introduced early in the stem cell field. (C) Correlation of hESC usage with time of derivation. Lines derived earlier are used more frequently than those lines derived later, a pattern that appears to be independent of policy influence. Error bars: standard deviation of mean.

Figure 2. A cumulative advantage model for global hESC usage patterns.

(A) Schematic of our simple cumulative advantage model. (B) The observed cumulative frequency distribution of hESC usage is in red; 1000 simulations of the model outlined in (A) are shown in grey. For illustrative purposes, the best fit of the model to the data is shown in black. (C) Policy-independent assumptions allow for the statistical detection of “anomalous” lines in the empirical data (all lines represented by circles). A predicted distribution year for each of the published 995 cell lines, based on their usage patterns, was computed and the predicted year was compared with the actual first publication year. A predicted distribution date strikingly earlier than the actual publication date indicates that a published line is more widely used than would be expected by comparison with those lines published around the same time. The green line separates from the bulk population those lines with a predicted distribution date at least 4 years earlier than their actual publication date. Seven hESC lines (z-scores in red) appear to have been distributed at least four years earlier than their actual publication date. Five of these lines (BG01, HSF6, CA1, KhES-3 and HS401; shown in grey) were in fact derived and distributed significantly before their first peer reviewed publication date (see Table S4). Lines used significantly more often than others published in the same year (z-score >2) are indicated in red. (D) Possible technological explanations for the observed unusual usage patterns of hESC lines HUES9 and WIBR3.

Figure 3. Worldwide usage patterns of hESC lines.

Changes over time in the number of studies using eligible (formerly NIH approved) and non-eligible hESC lines is shown. Panel (A) shows global patterns; (B) patterns in the US and; (C) patterns in CIRM funded studies. Although the Bush administration’s restrictions on hESC usage were lifted in March 2009 the number of publications exclusively using formerly restricted lines did not increase significantly in any region from 2009–2011. For a detailed statistical comparison of these data see Table S3.