A recent article from Scientists from Novartis and NextMove Software gives a very interesting overview of the reactions described in US patents over the last 40 years. The authors analysed over 200,000 US patents since 1976 and extracted over 1 million unique reactions. The data was analysed breaking down each patent into a larger number of reaction types, reaction yields, product properties over time, etc.
Similarly as in one of our previous blogs, this short summary is far from an exhaustive review of the article which is rich in data that should excite any organic chemists but an invitation to look more deeply in what medicinal chemists have been making these last 40 years.
The largest set of reactions extracted from the dataset is made of ‘heteroatom alkylations and arylations and acylation’, representing 27.8% of the classified reactions over 40 years, followed by ‘acylations and related processes’ (21.3% of the data) (Figure 1).
Looking more deeply into the data for the larger set of alkylations and arylations, some reactions such as the Williamson ether synthesis and chloro N-alkylations that were very popular in the mid 70s and 80s have fallen out of favour over the years. Similarly, as much of medicinal chemistry revolves around amide bond formation, the popular Schotten-Baumann acylation of the 70s and 80s has now been largely superseded by the advent of modern activated coupling reagents, now representing just over 50% of all acylation reactions (Figure 2).
The publication also covers the evolution of deprotections as a main class of reactions and how, since the mid 90s, the commercialisation of (Boc)2O has led to N-Boc as the nitrogen protecting group of choice. This reaction trend increase goes hand in hand with the sharp growth of amide bond formation depicted in Figure 2.
The analysis would not be correct if it did not capture the rise of the C-C bond formation and the ever increasing evolution toward a flatter molecular landscape. This class of reaction is very interesting as the sharp growth of C-C bond formation since the early 90s (Figure 1) is directly linked to the advent of transition metal catalysis and in particular to the Suzuki cross coupling reactions (Figure 3). Almost all non-Suzuki reaction type for the formation of C-C bonds since the 2000s have decreased in popularity, presumably linked to the increase in commercially available and stable boronic acids and extensive research efforts to improve on the Suzuki reliability and high yields.
Other trends are reviewed in the article, including a heatmap of the evolution of oxidations over time. The analysis clearly shows a move toward environmentally friendly oxidative reagents from the early 2000s, replacing the classic and highly toxic chromium-based Jones and Collins oxidations.
The authors have also analysed the yields of each reaction over time from the patent set. Overall, there is a clear decline in the median yield of all the reactions analysed. The authors associated this reduction of yields with the increasing popularity of flash purification and HPLC purification and library/array production to feed the ever increasing miniaturisation of screening techniques.
Finally, the authors also looked at the trend in the properties of novel compounds reported in US patents over the last 40 years (Figure 4). cLogP has climbed steadily over the last 40 years, peaking around 2005, whilst molecular weight has also been increasing and started to level off around 2010 at around 410 – 420 Da. Another clear trend of molecular properties of novel compounds over the last 40 years is the significant drop in rotatable bond count since the mid 90s, which is likely to be associated with the rise of the mighty Suzuki cross coupling reactions.
Figure 4. Evolution of physicochemical properties over time
The article reflects that although the modern medicinal chemist has a wide range of chemical reactions available to use in the design novel drugs, it seems that with the advent of modern acylation coupling reagents and palladium catalysed cross coupling, the modern medicinal chemist may be limiting himself to a very narrow range of reactions. The trend does not suggest this will change any time soon.
Blog written by Michael Paradowski