So, how has Google fared with some time–nearly 17 months–to marinate its patent search engine?

First off, Google hasn’t promoted the service out of beta yet.  I don’t know how Google decides to promote a new tool to production, but I have to think that many thousands of searches have been made with it since beta testing began, and they’ve had plenty of time to work out the kinks.  Beta schmeta…is it good enough to recommend?

Once again, I’ll compare Google Patent Search directly to the USPTO search engine, as Google has not yet added international patents. 

The Google search interface has been improved and provides simple form field searching for patent number, inventor, assignee, classification code (US and international), issue date, and filing date.  Google’s advanced search operators also work, including the “-” for NOT and “OR”.

The USPTO quick search allows search of all these fields plus some two dozen more, and their advanced search interface allows all of these delimiters to be accessed simultaneously, albeit using a complex syntax.  If you’re doing serious patent research, Google’s search interface will simply not provide you with enough search specificity to meet your needs.

For simple searches, the USPTO quick search and Google Patent advanced search interfaces are similar in user-friendliness and utility.  (The basic Google Patent search has very limited utility.)  So, the key differentiators will be result accuracy and result presentation.

Last time I looked at the Google service, it offered results in a list sorted by relevance-according-to-Google.  In my mind, this made the service virtually useless for industry-based searchers.  Google has since added in a date sorting.  I would have preferred having a patent number sorting option, but date at least makes the search worth doing.

As for relevance, I’m still scratching my head with Google.  It seems to return results in some quasi-random way.  For instance, search the exact phrase (in quotes) “Eli Lilly” the field assignee name, using the entire USPTO issued patent database (quick search) or Google Patent Search (advanced search).  The USPTO returns 3,793 hits dating back to 1971.  Google returns only 527 issued patents dated to 1976, and very strangely, only 461 filings of any status (issued plus applications combined–yes this number is lower than issued patnets alone!), and just 46 patent applications (USPTO returns 94 applications).

Clearly, there is something amiss with Google’s Patent Search.  It really does behave as a beta product, despite its protracted time in use.   It has some potential advantages to the USPTO search in theory, claiming, for example, full-content access to all US issued patents, whereas the USPTO offers full-text access only as far back as 1976.  But it remains fatally flawed for all research purposes except cherry-picking individual patents by number, when you are certain the patent number is correct.

In what must be the highest valuation paid for a Phase 2 drug and a pipeline of follow-ups (all in the same pharmacological category mind you), GSK is paying nearly three-quarters of a billion dollars for Sirtris. 

That’s $180 million for each of Sirtris’ four years of existence.

Okay, so Sirtris has its share of sirtuin-activator patent applications pending (at least 180, according to the company).  But those aren’t issued patents, are they? 

Three-quarters of a billion.  That’s roughly three-quarters of a billion for each of Sirtris’ one issued U.S. patent.

And they’ve got a small-molecule drug that apparently has cleared Phase 2.  But that leaves another development phase and a regulatory review to go before even getting to market by my math.  

Three-quarters of a billion with at least four more years of development time.

History tells me that pioneering small molecules for diabetes that have cleared this development hurdle have roughly a coin’s toss odds of eventually gaining major marketing approval, and a much lower chance of making it to market and becoming blockbusters.

And yet.

Three-quarters of a billion.  Roughly $9 milion per Sirtris employee.

Are you feeling valued yet?

Normally, I wouldn’t give a company so small in stature space in these pages, but this news release from tiny Symbollon (market cap pre-announcement circa $10M) stirred some memories for me.  You see Iogen, an oral formulation of molecular iodine–Symbollon’s entire pipeline–was also the first drug I was asked to evaluate for possible in-licensing.

That was back in 1998, or maybe 1999.  I remember feeling expertish.

At the time Symbollon was beginning its Phase 2 study of Iogen, but I believe they had some clinical data already in hand too. 

I remember being relatively unimpressed with the efficacy potential but also thinking: “Why not license the thing? It’s probably safe, there’s a chance it works, and it certainly will be cheap.”  Yeah…I was still pretty green then, not fully appreciating the substantial out-of-pocket and opportunity costs associated with any clinical-phase drug acquisition.  Other people more wizened than me realized the true costs, and my employer, also unimpressed with the efficacy data gathered at the time, took a pass.  As apprently did every other pharmaceutical company who saw the package. 

From a glance at the Symbollon website, it looks like the company went it alone for a while and even initiated the pivotal study referenced above by themselves.  Although it appears that they initiated the same study some six months later, after recruiting a license partner in Gardent Pharmaceuticals (nee Bioaccelerate Holdings).  Start a study twice?  Hey, why not?  Give investors their money worth I say.

I have no idea what became of Gardent, but they bailed on Iogen in 2006, leaving Symbollon to go it alone to the bitter end.  Symbollon had initially expected Phase 3 to wrap up in 12 months; it ended up taking nearly 40.  A lack of timely funds and a skeletal management team will do that.  Oh, and they blew through $20 million.  Not much to big pharma, but the opportunity cost would have been several times more.

Pity poor Symbollon, with it’s tiny market cap barely registering a blip and it’s hopes all faded.  My lesson finally complete.