In silico (i.e., computer simulation-based) research techniques are being increasingly utilized in new drug development, aided by advancements in computing power and technologies. In addition to drug target identification, in silico research programs now also screen for promising candidate compounds, predict side effects or drug resistance, and have proven to be particularly useful for drug design and optimization using large chemical libraries.

In silico data supporting patentability

In the above context, the extent to which in silico data can be relied upon to support a pharmaceutical invention has increasingly been the subject of debate. In order for pharmaceutical inventions that target the human body to meet specification description requirements, Korean courts and the Korean Intellectual Property Office ("KIPO") have generally required presentation of test examples with pharmacological data or equivalent descriptions that are sufficient to establish the pharmacological effect, under the rationale that it is generally difficult even for one skilled in the art to predict the effects of specific substances inside the human body. Exceptions to this rule are made only in special circumstances, such as if the underlying pharmacological mechanism of the pharmacological effect was already clearly established prior to the application filing date. 

Such test examples do not necessarily need to describe clinical trial results, as in vitro or animal test data are generally acceptable for description requirements purposes. However, according to KIPO's Examination Guidelines for biotechnology inventions, for an invention that predicts the interaction between a new drug candidate and a protein, or the therapeutic effect on a specific disease and the binding target of a substance, based only on in silico methods without confirmation through actual experiments, such inventions are not patentable as failing to meet utility or description requirements. In other words, even if a new drug candidate or its therapeutic effect is discovered through an in silico analysis method, actual experiments must still be conducted to confirm such therapeutic effects and described in the specification for an invention based on the drug to be patentable.

In silico data denying patentability (inventiveness)

On the other hand, the Korean IP High Court recently denied the inventiveness of a pharmaceutical use invention based on prior art which had predicted the pharmaceutical use from in silico data (IP High Court Decision 2021 Heo 5174, rendered on November 23, 2022). The Court reasoned that a person skilled in the art readily could have predicted the claimed pharmaceutical use from prior art documents disclosing a technique for predicting the therapeutic effect of a group of drug candidates in silico. 

The claimed invention in the case was related to a pharmaceutical use of raloxifene, a drug used to prevent and treat osteoporosis, as a COVID treatment, based on in vitro antiviral effect data against SARS-CoV-2. The prior art was a paper titled "Repurposing FDA-Approved Drugs for COVID-19 Using a Data-Driven Approach," which discussed raloxifene as a drug that could potentially be repurposed to treat COVID-19 based on connectivity mapping and molecular docking analysis (in silico techniques). The inventiveness of the claimed invention was denied both by KIPO and by the Intellectual Property Trial and Appeal Board (IPTAB) on appeal, and the applicant further appealed to the IP High Court.

The applicant primarily argued that the invention in question directly confirmed the effect of raloxifene in vitro, while the prior invention merely predicted the possible use of raloxifene for the treatment or prevention of SARS-CoV-2 infection through an in silico virtual search. Accordingly, the applicant asserted that the inventiveness of the claimed invention could not be denied because a person skilled in the art could not easily have predicted the invention from the prior art.

However, citing to the Supreme Court's prior ruling that "a pharmaceutical invention's inventiveness is denied if a person skilled in the art easily could have predicted the therapeutic effect of a specific substance on a specific disease from the prior art, in which case it cannot be considered that the therapeutic effect must be confirmed through clinical trials, etc., in the prior art," the IP High Court ruled that the claimed invention readily could have been predicted by a person skilled in the art from the prior art for the following reasons:

• The prior art disclosed the technical idea of selecting drug candidates in silico, and the connectivity mapping and molecular docking simulation used in the prior art corresponded to reliable in silico drug screening methods that had been widely used in the field of new drug development.
• It would have been obvious to one skilled in the art that: virus proliferation is inhibited when RdRp or Mpro functions within a cell are inhibited, so substances that inhibit RdRp or Mpro are related to the therapeutic effect against SARS-CoV-2 infection; in order to inhibit the RdRp or Mpro functions, drugs (ligands) must bind strongly to the specific receptors of the above enzymes; and by comparing the binding affinity between candidate SARS-CoV-2 infection treatment substances and receptors of RdRp or Mpro through molecular docking analysis, substances that can act more effectively on SARS-CoV-2 infection can be screened.
• From the disclosures of the prior art, it easily could have been inferred that a correlation existed between estrogen, a type of female hormone, and COVID-19 infection or related complications, and that the antiviral activity of raloxifene and tamoxifen was related to the estrogen regulation mechanism of these drugs. 
• There was no reason to believe in the superiority or inferiority of experimental evaluations or reliability among in silico, in vitro, and in vivo data, or that the tests must be conducted in the order of in silico, in vitro, and in vivo evaluations. Further, since in vitro methods do not directly observe reactions occurring inside living organisms, it could not be concluded that phenomena observed in vitro necessarily occurred in vivo, which is the same as in silico.

The above IP High Court decision means that even if a prior art reference discloses only in silico methods or data for a pharmacological effect and fails to meet description requirement, it can still serve as a basis for denying inventiveness if it is deemed that a person skilled in the art easily could have derived the invention from the prior art disclosure. This result seems to be due in part because in silico data is deemed less reliable than actual experimental data (in vivo, in vitro, etc.) when it comes to description requirements, whereas it is considered on more or less equal footing with in vivo or in vitro experimental data for determining inventiveness. As long as this remains the case, patent applicants in Korea should be vigilant of purely computer-based prior art, which may proliferate with continued advancements in computer technology. At the same time, patent applicants in Korea should be careful to generate and submit actual experimental results (e.g., in vivo, in vitro) to support pharmaceutical inventions discovered or developed using in silico techniques or data.