Emil W. Ciurczak, DoraMaxx Consulting11.20.19
Over the past few years, I have lauded PAT (Process Analytical Technology), QbD (Quality by Design), and, of course CM (Continuous Manufacturing) as means to producing faster, less expensive, and safer drug products. What hasn’t been emphasized is the environmental effects of “better” manufacturing. Environmental matters are a large arena, so I will attempt to break down the areas where process analyses and control will help. The fact is that “going green,” in many cases, also means saving big bucks.
Warehouse and production space
On any given production site, the largest footprint is from the warehouses. Warehouses for raw materials, intermediate products, final products, and packaged products, ready to ship. When I worked at Merck (1991-2), West Pont, PA, I was able to walk across campus for whatever purpose. When I visited in 2005, I needed to drive to get somewhere in a reasonable time. The reason? Mainly warehouses and some labs had been constructed.
I cannot directly address the cost per square foot of lighting and heating/air conditioning (HVAC), but assume we can cut at least 50% of our lighting and HVAC costs by not having to store intermediate (in-process) samples while waiting for either an analysis or piece of production equipment to be freed (cleaned and approved) for use in the next step of the process.
Loss from transfer
When the dry raw materials are transferred from storage and poured/scooped into a blender, some material is lost. When the blend is transferred to containers to another holding area to a granulator, more material is lost. During and after granulation (transfer), more is lost. In short, every time a powder blend is transferred from container to container a percentage is lost. When each process unit is washed, there is a volume of water, contaminated with detergent and API, washing down the drain.
Cost of traditional analysis
There are four major areas of expense: lab space, equipment, personnel, and chemicals. Let’s take a closer look at all four areas:
1. Laboratory space. When designing and expanding the lab space, we need to include the expenses of all utilities (water, electric, gasses, and HVAC). One of the most expensive rooms to construct in a Pharma company is the lab. Many meters of gas, vacuum, and water pipes are required, plus multiple electrical outlets per section often with voltage smoothing equipment, sinks, air hoods, and so forth, are also needed.
Also needed is the ancillary storage of solvents and chemicals in “explosion-proof” environments, which means more “brick and mortar” construction and ventilation. This means more space, electrical power, etc., all of which drive up costs.
2. Equipment. No matter which manner of analysis a company chooses for routine analysis, capital purchases must be made. For this example, I will just compare a typical HPLC units with an NIR-based on-line (commercially available) unit.
a. For the HPLC, a fully automated unit, capable of 30-50 samples over a 24-hour period, could cost between $35-50,000 (excluding columns, sample vials, etc.), ignoring IQ/OQ/PQ (necessary for any unit, often performed by the manufacturer). Operational costs are listed below.
b. For an in-line NIRS unit, the cost would be around $100,000. This unit would be capable of analyzing whole capsules or tablets—for ID of API, content uniformity, and prediction of dissolution times—at a rate of 100,000 units per hour.
c. Assuming a 3-5,000,000 tablet batch, the HPLC would be (destructively) measuring 0.007 to 0.004% of the batch for uniformity, while the NIR unit would be measuring (non-destructively) 100% of the batch using a blast of compressed air/nitrogen to eliminate any OOS sample.
d. One of the most under-estimated time costs is the length of time needed for an HPLC assay. For illustration:
i. Sample(s) must be gathered from production locale.
ii. Label and transport the samples to QC.
iii. Sign-in the samples and assign them to an analyst.
iv. The analyst then prepares standards, performs sample preparation (grind, dissolve, filter, pipette into vials).
v. Condition the HPLC, acquire computer files, place vials into unit and begin (15-20 min/sample).
vi. Supervisor checks chromatograms for errors, OK’s the results, assuming a repeat is not needed, and releases or fails the batch (“failure” may just mean repeating a larger number of samples).
vii. So, an HPLC assay without dissolution testing approaches one day in duration; the NIR time per sample is roughly 10 milliseconds, with no sample preparation.
3. Personnel. While the actual analytical instruments may be automated, they still need personnel to do the paperwork, sample preparation, solvent/mobile phase preparation, set-up and break-down of columns, etc. Under normal conditions, one analyst is assigned or uses one HPLC. The number of lots run per day per unit will vary on the make and model as well as the difficulty of sample preparation (often, up to one hour or more), so the number of LC units needed to service a busy company will vary. In HPLC, there are both operators and maintenance personnel, while for in/at-line analysis units, the technical personnel are both operators and maintainers of the hardware.
4. Chemicals, disposables. The cost of analysis can be estimated on an assay-by-assay basis:
a. Assuming a single HPLC assay takes:
i. 0.5 hour to obtain, label, and assign a sample.
ii. 0.5 to 1.0 hour to prepare the sample for injection, taking between 250mL to 500mL of solvent per sample, normally ~50% organic solvent.
iii. Between 10 to 30 minutes elution time per sample.
iv. At 1.0 to 1.5 mL/min of mobile phase, each sample could use 10 to 45mL of solvent, per sample.
b. The cost of one sample can then be estimated, if only for the expendables, ignoring HVAC and salaries of the people doing the work.
i. Assume that HPLC grade solvents may cost $25-$50 per liter.
ii. For a single sample, the cost could be (low vs. high)
1. 1.0mL/min x 10 min x $25/L x 1.0L/1000mL x 50% = $0.13 (low)
2. 1.5mL/min x 15 min x $50/L x 1.0L/1000mL x 50% = $1.13 (hi)
iii. If a content uniformity were performed (10 samples, pre- and post-standards), we can “spend” $1.13/sample x 22 samples (and standards) = $24.75 in solvent used.
iv. Add to that the cost and time of sample prep and the cost per sample increases even more.
v. And, unless times have changed, it costs more to (legally) dispose of the solvents than to purchase them.
c. Assume 50-60 lots of drug per week (with no re-runs or errors):
i. For the short/fast runs; $0.13 x 22 samples x 50 lots = $143 in solvents.
ii. For long/slower runs; $1.13 x 22 samples x 60 lots = $191.60 in solvents, alone.
iii. The cost of disposal must include storage, transport, paperwork, fees for storage at final site, etc. Aside from all steps having toxic chemicals and chance of spillage or human exposure, double the cost per assay.
d. The exact cost of filters, injection vials, and columns will vary, but could well equal the cost of solvents and may be more stressful on disposal sites than the actual solvents (which, in some cases may be recycled or burned).
e. While the exact number of calibrated glassware needed varies, the need for washing and storing them must be considered: washing requires water (cost plus usage), detergents, and drying (often heated). All use electricity, which is often generated from fossil fuel… not to mention burden on the sewage system.
Cost of quality
Clearly, something as drastic as a product recall would be financially disastrous—cost of product, transportation costs, off-the-shelf-time, reputation diminishing, fines from FDA—but would also exacerbate the environmental problems. If nothing else, the replacement batch would have the same losses as the recalled batch (process losses, transportation emissions, etc.) and the recalled batch would need to be safely destroyed and disposed. The disposed/destroyed lot also poses the same environmental problems as the rinse-water from equipment cleaning (or spillage of and/or improper handling of solvents), but on a much larger scale.
The API might be recovered prior to disposal, but that involves more solvents, which will eventually need to be disposed of or allowed to evaporate. The former has the potential to contaminate groundwater while the latter is a clean air hazard.
Observation and suggestion
Moving from “traditional” batch-style production to a simple PAT-mode will alone speed up production and minimize losses both financially and environmentally. Adding QbD will make for fewer OOS batches, avoiding batch destruction or recall. When, and if, real-time release (in-line measurements of final product) is added, the cost saving goes up by far more—zero recalled batches and a massive avoidance of lab-based assays.
More than a decade of accepted and successful submissions to the FDA and EMA have shown that end-of-process NIR (or Raman) monitoring can provide identification (verification of ID and level) of the API, API assay, content uniformity, and, for a bonus, dissolution predictions may be generated. All these save transportation (to the lab) time, solvents, columns, labor, etc.
When you are ready for the “big time,” continuous manufacturing is the “trifecta” of all these savings. There would be no transfer losses (with cleaning pollution), no lab solvent usage (destructive assays and labor costs), and real-time release. As an added attraction, the “footprint” of the process will be a mere fraction of batch processing and the warehouses need not be built. A win-win-win for the company, patient, and environment.
Emil W. Ciurczak
DoraMaxx Consulting
Emil W. Ciurczak has worked in the pharmaceutical industry since 1970 for companies that include Ciba-Geigy, Sandoz, Berlex, Merck, and Purdue Pharma, where he specialized in performing method development on most types of analytical equipment. In 1983, he introduced NIR spectroscopy to pharmaceutical applications, and is generally credited as one of the first to use process analytical technologies (PAT) in drug manufacturing and development.
Warehouse and production space
On any given production site, the largest footprint is from the warehouses. Warehouses for raw materials, intermediate products, final products, and packaged products, ready to ship. When I worked at Merck (1991-2), West Pont, PA, I was able to walk across campus for whatever purpose. When I visited in 2005, I needed to drive to get somewhere in a reasonable time. The reason? Mainly warehouses and some labs had been constructed.
I cannot directly address the cost per square foot of lighting and heating/air conditioning (HVAC), but assume we can cut at least 50% of our lighting and HVAC costs by not having to store intermediate (in-process) samples while waiting for either an analysis or piece of production equipment to be freed (cleaned and approved) for use in the next step of the process.
Loss from transfer
When the dry raw materials are transferred from storage and poured/scooped into a blender, some material is lost. When the blend is transferred to containers to another holding area to a granulator, more material is lost. During and after granulation (transfer), more is lost. In short, every time a powder blend is transferred from container to container a percentage is lost. When each process unit is washed, there is a volume of water, contaminated with detergent and API, washing down the drain.
Cost of traditional analysis
There are four major areas of expense: lab space, equipment, personnel, and chemicals. Let’s take a closer look at all four areas:
1. Laboratory space. When designing and expanding the lab space, we need to include the expenses of all utilities (water, electric, gasses, and HVAC). One of the most expensive rooms to construct in a Pharma company is the lab. Many meters of gas, vacuum, and water pipes are required, plus multiple electrical outlets per section often with voltage smoothing equipment, sinks, air hoods, and so forth, are also needed.
Also needed is the ancillary storage of solvents and chemicals in “explosion-proof” environments, which means more “brick and mortar” construction and ventilation. This means more space, electrical power, etc., all of which drive up costs.
2. Equipment. No matter which manner of analysis a company chooses for routine analysis, capital purchases must be made. For this example, I will just compare a typical HPLC units with an NIR-based on-line (commercially available) unit.
a. For the HPLC, a fully automated unit, capable of 30-50 samples over a 24-hour period, could cost between $35-50,000 (excluding columns, sample vials, etc.), ignoring IQ/OQ/PQ (necessary for any unit, often performed by the manufacturer). Operational costs are listed below.
b. For an in-line NIRS unit, the cost would be around $100,000. This unit would be capable of analyzing whole capsules or tablets—for ID of API, content uniformity, and prediction of dissolution times—at a rate of 100,000 units per hour.
c. Assuming a 3-5,000,000 tablet batch, the HPLC would be (destructively) measuring 0.007 to 0.004% of the batch for uniformity, while the NIR unit would be measuring (non-destructively) 100% of the batch using a blast of compressed air/nitrogen to eliminate any OOS sample.
d. One of the most under-estimated time costs is the length of time needed for an HPLC assay. For illustration:
i. Sample(s) must be gathered from production locale.
ii. Label and transport the samples to QC.
iii. Sign-in the samples and assign them to an analyst.
iv. The analyst then prepares standards, performs sample preparation (grind, dissolve, filter, pipette into vials).
v. Condition the HPLC, acquire computer files, place vials into unit and begin (15-20 min/sample).
vi. Supervisor checks chromatograms for errors, OK’s the results, assuming a repeat is not needed, and releases or fails the batch (“failure” may just mean repeating a larger number of samples).
vii. So, an HPLC assay without dissolution testing approaches one day in duration; the NIR time per sample is roughly 10 milliseconds, with no sample preparation.
3. Personnel. While the actual analytical instruments may be automated, they still need personnel to do the paperwork, sample preparation, solvent/mobile phase preparation, set-up and break-down of columns, etc. Under normal conditions, one analyst is assigned or uses one HPLC. The number of lots run per day per unit will vary on the make and model as well as the difficulty of sample preparation (often, up to one hour or more), so the number of LC units needed to service a busy company will vary. In HPLC, there are both operators and maintenance personnel, while for in/at-line analysis units, the technical personnel are both operators and maintainers of the hardware.
4. Chemicals, disposables. The cost of analysis can be estimated on an assay-by-assay basis:
a. Assuming a single HPLC assay takes:
i. 0.5 hour to obtain, label, and assign a sample.
ii. 0.5 to 1.0 hour to prepare the sample for injection, taking between 250mL to 500mL of solvent per sample, normally ~50% organic solvent.
iii. Between 10 to 30 minutes elution time per sample.
iv. At 1.0 to 1.5 mL/min of mobile phase, each sample could use 10 to 45mL of solvent, per sample.
b. The cost of one sample can then be estimated, if only for the expendables, ignoring HVAC and salaries of the people doing the work.
i. Assume that HPLC grade solvents may cost $25-$50 per liter.
ii. For a single sample, the cost could be (low vs. high)
1. 1.0mL/min x 10 min x $25/L x 1.0L/1000mL x 50% = $0.13 (low)
2. 1.5mL/min x 15 min x $50/L x 1.0L/1000mL x 50% = $1.13 (hi)
iii. If a content uniformity were performed (10 samples, pre- and post-standards), we can “spend” $1.13/sample x 22 samples (and standards) = $24.75 in solvent used.
iv. Add to that the cost and time of sample prep and the cost per sample increases even more.
v. And, unless times have changed, it costs more to (legally) dispose of the solvents than to purchase them.
c. Assume 50-60 lots of drug per week (with no re-runs or errors):
i. For the short/fast runs; $0.13 x 22 samples x 50 lots = $143 in solvents.
ii. For long/slower runs; $1.13 x 22 samples x 60 lots = $191.60 in solvents, alone.
iii. The cost of disposal must include storage, transport, paperwork, fees for storage at final site, etc. Aside from all steps having toxic chemicals and chance of spillage or human exposure, double the cost per assay.
d. The exact cost of filters, injection vials, and columns will vary, but could well equal the cost of solvents and may be more stressful on disposal sites than the actual solvents (which, in some cases may be recycled or burned).
e. While the exact number of calibrated glassware needed varies, the need for washing and storing them must be considered: washing requires water (cost plus usage), detergents, and drying (often heated). All use electricity, which is often generated from fossil fuel… not to mention burden on the sewage system.
Cost of quality
Clearly, something as drastic as a product recall would be financially disastrous—cost of product, transportation costs, off-the-shelf-time, reputation diminishing, fines from FDA—but would also exacerbate the environmental problems. If nothing else, the replacement batch would have the same losses as the recalled batch (process losses, transportation emissions, etc.) and the recalled batch would need to be safely destroyed and disposed. The disposed/destroyed lot also poses the same environmental problems as the rinse-water from equipment cleaning (or spillage of and/or improper handling of solvents), but on a much larger scale.
The API might be recovered prior to disposal, but that involves more solvents, which will eventually need to be disposed of or allowed to evaporate. The former has the potential to contaminate groundwater while the latter is a clean air hazard.
Observation and suggestion
Moving from “traditional” batch-style production to a simple PAT-mode will alone speed up production and minimize losses both financially and environmentally. Adding QbD will make for fewer OOS batches, avoiding batch destruction or recall. When, and if, real-time release (in-line measurements of final product) is added, the cost saving goes up by far more—zero recalled batches and a massive avoidance of lab-based assays.
More than a decade of accepted and successful submissions to the FDA and EMA have shown that end-of-process NIR (or Raman) monitoring can provide identification (verification of ID and level) of the API, API assay, content uniformity, and, for a bonus, dissolution predictions may be generated. All these save transportation (to the lab) time, solvents, columns, labor, etc.
When you are ready for the “big time,” continuous manufacturing is the “trifecta” of all these savings. There would be no transfer losses (with cleaning pollution), no lab solvent usage (destructive assays and labor costs), and real-time release. As an added attraction, the “footprint” of the process will be a mere fraction of batch processing and the warehouses need not be built. A win-win-win for the company, patient, and environment.
Emil W. Ciurczak
DoraMaxx Consulting
Emil W. Ciurczak has worked in the pharmaceutical industry since 1970 for companies that include Ciba-Geigy, Sandoz, Berlex, Merck, and Purdue Pharma, where he specialized in performing method development on most types of analytical equipment. In 1983, he introduced NIR spectroscopy to pharmaceutical applications, and is generally credited as one of the first to use process analytical technologies (PAT) in drug manufacturing and development.
