The number of observations and descriptive statistics for each dataset are presented in Tables 1 and Fig. 2, respectively.
Table 1
Descriptive statistics of variables in each dataset
Variable | Category | First-in-class drugs | Follow-on drugs | Total |
N | (%) | N | (%) | N |
Molecule Type | Chemical | 113 | (28.3) | 216 | (12.6) | 329 |
Biologics (Non-antibody) | 29 | (14.6) | 34 | (11.9) | 63 |
Antibody Biologics | 56 | (57.1) | 36 | (75.5) | 92 |
Orphan Drug | Yes | 119 | (60.1) | 105 | (36.7) | 224 |
No | 79 | (39.9) | 181 | (63.3) | 260 |
Disease | Oncology | 54 | (27.3) | 89 | (31.1) | 143 |
Non-oncology | 144 | (72.7) | 197 | (68.9) | 341 |
Pharmacological Action | Activator | 39 | (19.7) | 59 | (20.6) | 98 |
Suppressor | 133 | (67.2) | 189 | (66.1) | 322 |
Others | 26 | (13.1) | 38 | (13.3) | 64 |
Drug Target | Protein | 98 | (49.5) | 113 | (39.5) | 211 |
Ion channel | 5 | (0.5) | 10 | (3.5) | 15 |
Ligand | 11 | (5.6) | 20 | (7.0) | 31 |
Receptor | 67 | (33.8) | 110 | (38.5) | 177 |
Others | 17 | (8.6) | 33 | (11.5) | 50 |
Firm Size | Top 20 | 82 | (41.4) | 114 | (39.9) | 196 |
Others | 116 | (58.6) | 172 | (60.1) | 288 |
Country | US | 132 | (66.7) | 157 | (54.9) | 289 |
Non-US | 66 | (33.3) | 129 | (45.1) | 195 |
Development Strategy | In-licensed | 64 | (32.3) | 98 | (34.3) | 162 |
Acquisition | 48 | (24.3) | 55 | (19.2) | 103 |
Self-originated | 86 | (43.4) | 133 | (46.5) | 219 |
(Insert Table 1 Here)
(Insert Fig. 2 Here)
Among the 484 NMEs from 2010 to 2022, 329 were chemicals and 113 were first-in-class drugs. Of the 63 biologics (excluding antibody biologics), 29 were first-in-class. Of the 92 antibody biologics, 56 (60.9%) were first-in-class. Among the first-in-class drugs, 56 (28.3%) were antibody biologics, exceeding the 36 (12.6%) antibody biologics among follow-on drugs.
Of the 484 NMEs, 224 were orphan drugs. Among first-in-class drugs, 119 (60.1%) were orphan drugs, exceeding the 105 (36.7%) orphan drugs among follow-on drugs. There were 143 oncology drugs. Among the first-in-class drugs, 54 (27.3%) were oncology drugs and 114 (72.7%) were non-oncology drugs. Regarding drug targets, 211 classified as were proteins, 177 as receptors, 31 as ligands, 15 as ion channels, and 50 as others. Among the drug targets for first-in-class drugs, 98 (49.5%) were categorized as proteins, 67 (33.8%) as receptors, 11 (5.6%) as ligands, 5 (0.5%) as ion channels, and 17 (8.6%) as others.
From 2010 to 2022, of the 484 NME license granted by the US FDA, 289 were for US companies and 195 were for non-US companies, based on their headquarters. US companies accounted for 66.7% (132) of the first-in-class drugs and 54.9% (157) of the approved follow-on drugs. In the analysis of the combination of country and development strategy, the 123 US and self-originated NMEs comprised 60 (48.8%) first-in-class drugs and the 96 non-US and self-originated NMEs comprised 26 (27.1%) first-in-class drugs.
A model that affects the approval of first-in-class drugs from the US FDA was constructed and analyzed using Stata 16 and logistic regression. Each analysis yielded a standard error (SE) that was robust to heteroscedasticity. In addition, the VIF was calculated to confirm multicollinearity; this was not a concern for any model.
(Insert Table 2 Here)
Table 2
Results of logistic regression analysis
Category | OR | Robust SE | 95% CI | P-value |
Molecule Type | | | | |
Chemical | Reference | | | |
Biologics (Non-antibody) | 1.173 | 0.390 | 0.612–2.250 | 0.630 |
Antibody Biologics | 3.084 | 0.924 | 1.715–5.548 | 0.000** |
Orphan Drug | | | | |
No | Reference | | | |
Yes | 3.581 | 0.850 | 2.249–5.701 | 0.000** |
Disease | | | | |
Non-oncology | Reference | | | |
Oncology | 0.359 | 0.100 | 0.208–0.619 | 0.000** |
Pharmacological Action | | | | |
Others | Reference | | | |
Activatorsa | 0.869 | 0.342 | 0.402–1.879 | 0.722 |
Suppressorsb | 0.904 | 0.303 | 0.469–1.745 | 0.764 |
Drug Target | | | | |
Others | Reference | | | |
Protein | 2.262 | 0.920 | 1.020–5.018 | 0.045* |
Ion Channel | 1.151 | 0.840 | 0.275–4.816 | 0.847 |
Ligand | 1.126 | 0.639 | 0.370–3.427 | 0.834 |
Receptor | 1.961 | 0.823 | 0.861–4.465 | 0.109 |
Firm Size | | | | |
Others | Reference | | | |
Top 20 | 1.232 | 0.274 | 0.796–1.905 | 0.349 |
Strategy and Country | | | | |
Non-US & Self-originated | Reference | | | |
Non-US & In-licensed | 1.118 | 0.447 | 0.510–2.449 | 0.780 |
Non-US & Acquisition | 2.260 | 0.901 | 1.035–4.938 | 0.041* |
US & Self-originated | 2.680 | 0.822 | 1.469–4.889 | 0.001* |
US & In-Licensed | 1.996 | 0.637 | 1.068–3.731 | 0.030* |
US & Acquisition | 2.118 | 0.823 | 0.988–4.538 | 0.054 |
Constant | 0.118 | 0.049 | 0.052–0.266 | 0.000 |
Pseudo R2 | 0.1155 | | | |
Mean VIF | 1.87 | | | |
Hosmer–Lemeshow | x2 = 5.18 (p = 0.738) | | |
aAgonist and stimulator. |
bAntagonist and inhibitor. |
*p < 0.05, **p < 0.01. |
The results of logistic regression showed that antibody biologics (odds ratio (OR) = 3.084, p < 0.01), orphan drugs (OR = 3.581, p < 0.01), oncology drugs (OR = 0.359, p < 0.01), protein targets (OR = 2.262, p < 0.05), self-originated strategy by US companies (OR = 2.680, p < 0.01), in-licensed by US companies (OR = 1.996, p < 0.05), and acquisition strategy by non-US companies (OR = 2.260, p < 0.05) were statistically significant factors in the FDA’s approval of first-in-class drugs (Table 2). The results indicate that the probability of the approval of first-in-class drugs was 3.084 times higher for antibody biologics than that for chemical drugs, that for orphan drugs was 3.581 times higher than that for nonorphan drugs, and that for oncology drugs was 0.359 times lower than that for non-oncology drugs. The probability of the approval of first-in-class new drugs that targeted proteins was 2.262 times higher than that of other targets.
The probability of the approval of first-in-class of new drugs developed through self-originated strategy by US pharmaceutical companies, in-licensed strategy by US companies, and acquisition strategy by non-US companies is 2.680, 1.996, and 2.260 higher, respectively, than those developed through the self-originated strategy by non-US companies. The probability of the first-in-class approval of new drugs developing other variables was not observed to have a significant effect.
Factors that influenced the US FDA approval of first-in-class drugs were analyzed using logistic regression in two different periods: 2011–2016 and 2017–2022 (Table 3). To perform data analysis over a 6-year period, data from 2010 were excluded. We divided the time frames into the two periods around 2016 because of certain changes in the US FDA in 2016. In 2016, the US FDA’s approval of new drugs was the lowest since 2007; on 13 December 2016, the 21st Century Cure Act was signed into law, and included measures intended to streamline drug approvals.
Table 3
Results of logistic regression analysis for 2011–2016 vs. 2017–2022
| 2011–2016 | 2017–2022 |
Category | OR (95% CI) | P-value | OR (95% CI) | P-value |
Molecule Type | | | | |
Chemical | Reference | | | |
Biologics (Non-antibody) | 2.370 (0.691–8.126) | 0.170 | 1.002 (0.404–2.486) | 0.996 |
Antibody Biologics | 3.138 (1.005–9.799) | 0.049* | 3.123 (1.447–6.742) | 0.004** |
Orphan Drug | | | | |
No | Reference | | | |
Yes | 3.252 (1.377–7.677) | 0.007** | 3.997 (2.103–7.597) | 0.000** |
Disease | | | | |
Non-oncology | Reference | | | |
Oncology | 0.424 (0.156–1.155) | 0.093 | 0.279 (0.135–0.576) | 0.001** |
Action | | | | |
Others | Reference | | | |
Activatorsa | 1.398 (0.397–4.923) | 0.602 | 0.799 (0.266–2.402) | 0.689 |
Suppressorsb | 1.957 (0.662–5.788) | 0.225 | 0.597 (0.235–1.515) | 0.277 |
Drug Target | | | | |
Others | Reference | | | |
Protein | 0.901 (0.271–2.992) | 0.864 | 4.444 (1.382–14.292) | 0.012* |
Ion Channel | 2.667 (0.526–13.512) | 0.236 | 1.000 (omitted) | |
Ligand | 0.190 (0.029–1.242) | 0.083 | 3.530 (0.765–16.287) | 0.106 |
Receptor | 0.946 (0.267–3.358) | 0.932 | 2.952 (0.901–9.670) | 0.074 |
Firm Size | | | | |
Others | Reference | | | |
Top 20 | 1.197 (0.568–2.523) | 0.636 | 1.133 (0.609–2.109) | 0.693 |
Strategy and Country | | | | |
Non-US & Self-originated | Reference | | | |
Non-US & In-licensed | 1.430 (0.384–5.325) | 0.594 | 1.146 (0.397–3.311) | 0.801 |
Non-US & Acquisition | 2.641 (0.602–11.586) | 0.198 | 3.669 (1.165–11.562) | 0.026* |
US & Self-originated | 2.051 (0.757–5.561) | 0.158 | 4.090 (1.667–10.032) | 0.002** |
US & In-Licensed | 1.899 (0.700–5.147) | 0.208 | 2.341 (0.916–5.980) | 0.075 |
US & Acquisition | 2.646 (0.754–9.287) | 0.129 | 1.731 (0.629–4.762) | 0.288 |
Constant | 0.141 | 0.003 | 0.082 | 0.000 |
Pseudo R2 | 0.1216 | | 0.1540 | |
Mean VIF | 2.10 | | 1.87 | |
Hosmer–Lemeshow | x2 = 7.34 (p = 0.501) | | x2 = 6.37 (p = 0.606) | |
aAgonist and stimulator. |
bAntagonist and inhibitor. |
*p < 0.05, **p < 0.01. |
(Insert Table 3 Here)
Following the test of the significance of the regression coefficient, antibody biologics were observed to have a high probability of approval with an OR = 3.138 (p < 0.05) and 3.123 (p < 0.01). In addition, in both periods, the probability of the first-in-class approval of orphan drugs was high at OR = 3.252, 3.997, and significant at p < 0.01. Whether the new drugs were oncology drugs was not significant from 2011 to 2016, but was significant from 2017 to 2022 (OR = 0.279, p < 0.01). This means that over time, new anticancer drugs might have a negative impact on first-in-class designation. The selection of the drug target as protein was not significant between 2011 and 2016, but was significant (OR = 4.444, p < 0.05) between 2016 and 2022. Recently, protein targets have been considered as factors that influence innovative drugs.
For the corporation-based variable, the country of the applicant and development strategy of new drugs changed over time. The development of new drugs by US pharmaceutical companies through the self-originated strategy and by non-US companies through acquisition strategy were significant between 2017 and 2022, with OR = 3.669 (p < 0.05) and 4.090 (p < 0.01), respectively, but not between 2011 and 2016. From the above results, the factors influencing the first-in-class drugs approved by the US FDA have changed over time.