How many diabetics could be served by a small-scale insulin production system?

7 min read

Input is welcome! Let’s crowdsource this!

Open Insulin has proposed small-scale, localized production of insulin as a way to address large-scale, centralized manufacturing that contributes to oligopolic control over insulin production and access. We get asked how many people this type of small-scale system could serve. This post is an attempt to answer that question. Below you’ll find the calculations supporting the answer in increasing levels of detail.

Question: How many people could be served by a small-scale production system, something that could fit on top of a lab bench or table?

Answer: An average of 13,685 diabetics

Detail I: high level

The number of people with diabetes that could be served varies greatly based on a number of factors. The range is from about 4,400 to 45,000, with an average of 13,685 people. .

a diagram describing the below estimations

Low estimation:  low yield / high consumption = 16.87g/0.0038g = 4,439 diabetics

High estimation:  high yield / low consumption = 50.18g/0.0011g = 45,618 diabetics

Detail II: more detail

Consumption formula

Insulin consumption among insulin-dependent diabetics ranges from 30 to 100 units per day. Assuming a conversion rate of 1g = 26,000 units, people who require insulin consume between 1.1 to 3.8 mg of insulin per day, with an average of 2.45 mg.

Production formula

a diagram describing the below purification formula


1. The yield for Pichia pastoris strain ranges from  1.5 – 3.9 g/L of insulin precursor (IP) utilizing a 15L bioreactor (Mansur 2005, Polez 2016, Gurramkonda 2010 and Wang 2018).

2. This size yield has been tested for small bioreactors from 7 to 15 liters. As production is scaled up to larger units, yield estimates change; in other words, our estimate is specific to small bioreactors and does not apply readily for larger manufacturing equipment.

3. The purification part takes 1 day, while the production part takes an average of 6 days. We could have up to 6 bioreactors to optimize production.

4. The purification of insulin involves chromatography followed by the cleavage of the IP to get insulin. The variation in yield can be due to the quality of the column, the skills of the scientist and the quality of the batch.

Let’s play! Move around the cursor to choose different values for the parameters and look at how it will impact the final result! Can you guess the size of the city you will be able to serve with your set up?

Detail III: even more detail!

The Open Insulin Project is developing a protocol using the yeast strain Pichia pastoris as a host and combining chromatography techniques to purify the insulin precursor (IP). In the literature, the yield for the IP ranges from 0.25 g L-1  (Mansur 2005), 2.26 g L-1 (Polez 2016), 3.84 g L-1 (Gurramkonda 2010) to 6.69 g L-1  (Wang 2018). These yields have been obtained for small bioreactors (7-15L). Taking into account the technological improvements in the last decade, we can expect to get a yield between 2.26 to 6.69  g L-1.

The IP is secreted and can be found in the culture medium. The purity of IP in the supernatant is low and we will use different chromatography methods to isolate the IP. The next step will be to perform transpeptidation to obtain insulin. A crystallization step can be added to increase the purity of the sample. Overall, the process from culture supernatant to pure insulin can take up to 3 days and only 50% of the insulin will be recovered.

The maximum protein production in yeast is reached after 6 days when the bioreactor is cleared out and the supernatant harvested for purification. The purification process contains 3 main steps using different equipment and lasting less than a day. For optimal use of the equipment, 6 small bioreactors (7-15 L) can work simultaneously, with one bioreactor cleared out per day.  The advantage to working with multiple bioreactors and not a large one is to be more resilient against contamination or technical problems. If one bioreactor goes down, for example, the others will keep up production. After the first 6 days of set up, the small-scale production system will be able to produce a batch per day, that is, a yield between 2.26 to 6.69  g L-1 of purified IP that is then converted to insulin.

Now, pure insulin can be sent for formulation (including sterilization and packaging) that can take up to 2 weeks. In the meantime, a sample of the batch will be sent for stability analysis and activity tests (e.g. ELISA). Activity tests are important as they will determine the number of units (active insulin) per ml. The conversation rate standard, established by the World Health Organization (Heinemann 2010), between insulin mass and insulin unit is around 1g = 26,000 units but can vary from one batch to another. The stability test will make sure that the insulin is stable over a month at room temperature.

Assuming 24/7 operation, the small-scale production system will be able to produce on average 33.53 g of pure insulin per day. Depending on the yield obtained with the modified host we are currently developing, the yield of pure insulin could range from 16.87g to 50.18g.

To understand how many diabetics could be served, we calculated the insulin unit need per day taking into account the variability of insulin consumption (~30 to 100 units). Insulin consumption varies from patient to patient, is proportional to weight and also dependent on the injection method (syringe or loop). If we assume the conversion rate from 1 gram to 26,000 units (Heinemann 2010), diabetics consume between 1.1 to 3.8 mg of insulin per day, with an average of 2.45 mg.

Let’s now run these numbers again:

Avg  estimation:  avg  yield / avg  consumption =33.53g/0.00245g = 13,685 diabetics

Low estimation:  low yield / high consumption = 16.87g/0.0038g = 4,439 diabetics

High estimation:  high yield / low consumption = 50.18g/0.0011g = 45,618 diabetics

These estimations don’t take into account the potential loss of some of the insulin during formulation as well as the sample needed for analysis. This shouldn’t be more than 10% of the total amount of insulin produced. We also assumed 24/7 operation. Depending on the human labor available, it may be less. Some batches may be contaminated or have bad quality. Another factor to take into account is the differences in yield between different forms of insulin, such as insulin glargine versus lispro. Overall, the low estimation is probably a good estimation of what can be done.

Nicole Foti & Louise Lassalle


Polez S, Origi D, Zahariev S, Guarnaccia C, Tisminetzky SG, et al. (2016) A Simplified and Efficient Process for Insulin Production in Pichia pastoris. PLOS ONE 11(12): e0167207.

Nabih A Baeshen, Mohammed N Baeshen, Abdullah Sheikh, Roop S Bora, Mohamed Morsi M Ahmed, Hassan A I Ramadan, Kulvinder Singh Saini & Elrashdy M Redwan (2014) Cell factories for insulin production. Microbial Cell Factories volume 13, Article number: 141

Mansur, M., Cabello, C., Hernández, L. et al. Biotechnol Lett (2005) 27: 339.

Chandrasekhar Gurramkonda, Sulena Polez, Natasa Skoko, Ahmad Adnan, Thomas Gäbel, Dipti Chugh, Sathyamangalam Swaminathan, Navin Khanna, Sergio Tisminetzky & Ursula Rinas (2010) Application of simple fed-batch technique to high-level secretory production of insulin precursor using Pichia pastoris with subsequent purification and conversion to human insulin. Microbial Cell Factoriesvolume 9, Article number: 31

Wang, J., Wang, X., Shi, L. et al. J Ind Microbiol Biotechnol (2018) 45: 25.

Lutz Heinemann. Insulin Assay Standardization: Leading to Measures of Insulin Sensitivity and Secretion for Practical Clinical Care. Diabetes Care 2010 Jun; 33(6): e83-e83.

“World Health Organization insulin standard from 1987 has a potency of 26,000 units/g”

Ask: seeking a volunteer to build interactive calculator to embedded in our website