Three new drugs that could fight type 2 diabetes – what can patent analytics tell us?

Type 2 diabetes (too much sugar in the blood) is a disease that is thought to affect 360 million people world wide, and can shorten life expectancy by 10 years. For this reason the concept of a drug that may help reduce Type 2 diabetes by increasing the amount of sugar removed from the body with urine has great promise, and early results suggest this can reduce mortality, blood sugar, blood pressure and even help with weight loss.

This drug is “Jardiance” (technical name of empagliflozin) and known to be a “SGLT2 inhibitor”. This is the third known such inhibitor, along with “Invokana” (canagliflozin) and “Forxiga” (dapagliflozin). 

Ambercite are not medical doctors and so suggest that you consult your doctor for advice on whether any of these drugs would work for you or not.  

But we are patent analysts, and so were intrigued on what the patent analytics would look like for these three potentially very valuable drugs.

When looking at this sorts of analyses, the first issue is identifying the patents involved. The likes of Orange Book and similar sites can help identify the patents underpinnign commercially available drugs, and is so so often the case, each of these drugs was protected by multiple patents, with mutlitple filing and expiry dates. However to make things manageable we have chosen to simply analyse the first granted US patent for each drug. In a commercial assessment you would of course look at all of the family members, but this is intended to make things manageable.

And how did we analyse these three patents? We used our product Cluster Searching. This has a number of features, and most important for this analysis, these includes

  1. We assign to every patent in our database an AmberScore value, which is a general predictor of patent importance. Amberscore has been normalised so that the average AmberScore value for granted US patents in 1
  2. We assess the Total Licensing Potential for each patent. This is a simple predictor of the relative commercial importance of each patent. 

This is best shown in an example. We identified that the first granted US patent for empagliflozin was US6303661, and added this number into Cluster Searching. 




When we ran this query, the Patents searched part of our website showed  that;



So the patent was originally filed by Priobiodrug on 6th October 1998, and has an AmberScore value of 27.51. or more than 27 times as high as the average US patent. This is a good patent, but I suspect the owners know this already.

Arrmed with this information, we can reset the filing date filter in the query area



Which in turn will suggest the most similar prior art patents. In this case, this is US5939560, filed by Ferring BV in 1996 – and some others are shown below. Note that the 4th column shown in our “Similarity metric” – a higher value suggests greater similarity. 



From a commercial viewpoint, perhaps the patents that followed on from this are the most interesting. To find this, we reset the date filter so that it is showing patents filed after this filing date



This will produce a different set of reults, with the top three shown below



You will note from these results several things

  • The similarity values range from 59 downwards. We suggest that any similarity score of 2 or more is worth noting, so that there are a number of similar patents in these
  • We supply some values in the “L.P.” column. This is “LIcensing Potential”, which is a combination of the similarity and AmberScore of each patent in the list. 
  • The Citation types (second to last column) include a combination of Known (direct forward citaitons) and “Unknown” citations – the latter includes patents which would not be picked up in a general search, but which we still believe to be relevant
  • The list of owners include both related parties to the patent owner (Priobiodrug, the original applicant) and unrelated parties (Novarits)
  • You can export the results to Excel

From this table, we can now compute a total “Net Licensing Potential” – which we could define as

Net Licensing Potential = The sum of the Licensing Potential values for patents

  • filed after the original filing date of a patent, and
  • where the owner is not the original party or related party, and
  • where there is a minimum similarity score of 2


For this patent, there are a lot of patents in this list that meet these criteria – 512 in fact, and we can calculate the Net Licensing Potential”to be 2329, an exceptionally high value. 


Having done all of this analysis, it is very easy to repeat this for the other two drugs listed. The results are summarised below – we have also copied the expiry date  from the listing of this patent in OrangeBook, athough of  course these drugs are likely to be protected by multiple patents with different expiry dates. 


Brand name




Technical name




Earliest granted US patent number


US 6414126

US 7943788

Priority date

24 Apr 1997

12 Oct 1999

1 Aug 2003

Most similar prior art patent

US5939560, filed by Ferring BV in 1996

WO1998031697, filed by Sankyo Company Limited in 1998

US7094763, filed by  Janssen Pharmaceuticals in 2004

Known patent applicants / licensees

Probiodrug, Prosidion / Boehringer Ingelheim, Eli Lilly and Company

AstraZeneca, Bristol-Myers Squibb

Mitsubishi Tanabe Pharma, Janssen Pharmaceutical

Expiry date for listed patent

24 Apr 2017

4 Oct 2020

14 Jul 2027





Net Licensing Potential




Leading owners of similar and later patents – Total Licensing Potential in forward citation network)

Merck – 208

Novartis – 185

Takeda Pharma – 185

Boehringer Ingelheim – 580

Kissei Pharma – 138

Janssen Pharma – 94

Boehringer Ingelheim – 363

Theracos – 645

Kissei Pharma – 43


This table suggests that empagliflozin has the highest scoring patent. Of note, it was the first patent filed.


What we have we learnt about patent analytics in general?

In this case study, to some extent, we have cheated by looking out for patents for likely to be very important drugs. Putting this aside, we can see that it is possible to quantify the potential value of any patent using these patent analytics tools. We can imagine many applications for this capability, including

  • portfolio reviews
  • aqusition due diligence
  • patent valuation

These processse can be done quickly and efficiently, and so provide a robust means of assessing patent and IP value. And indeed Cluster Searching is being used all of the time for these sorts of analyses. 

Read More