Stability profiles of drug products extended beyond labeled expiration dates

Stability Profiles of Drug Products Extendedbeyond Labeled Expiration Dates ROBBE C. LYON,1 JEB S. TAYLOR,1 DONNA A. PORTER,2 HULLAHALLI R. PRASANNA,1 AJAZ S. HUSSAIN3 1Division of Product Quality Research, Center for Drug Evaluation and Research, Food and Drug Administration,HFD-941, White Oak, Life Sciences Building 64, 10903 New Hampshire Avenue, Silver Spring, Maryland 20993-0002 2Division of Field Science, Office of Regional Operations, Office of Regulatory Affairs, Food and Drug Administration,Rockville, Maryland 20857 3Vice President & Global Head of Biopharmaceutical Development, Sandoz, 506 Carnegie Center, Princeton,New Jersey 08540 Received 6 January 2006; accepted 17 March 2006 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20636 ABSTRACT: The American Medical Association has questioned whether expirationdating markedly underestimates the actual shelf life of drug products. Results from theshelf life extension program (SLEP) have been evaluated to provide extensive data toaddress this issue. The SLEP has been administered by the Food and DrugAdministration for the United States Department of Defense (DOD) for 20 years. Thisprogram probably contains the most extensive source of pharmaceutical stability dataextant. This report summarizes extended stability profiles for 122 different drug products(3005 different lots). The drug products were categorized into five groups based onincidence of initial extension failures and termination failures (extended lot eventuallyfailed upon re-testing). Based on testing and stability assessment, 88% of the lots wereextended at least 1 year beyond their original expiration date for an average extension of66 months, but the additional stability period was highly variable. The SLEP datasupports the assertion that many drug products, if properly stored, can be extended pastthe expiration date. Due to the lot-to-lot variability, the stability and quality of extendeddrug products can only be assured by periodic testing and systematic evaluationof each lot. ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci95:1549–1560, 2006Keywords: shelf life; drug stability; shelf life extension program; SLEP; expiration than their labeled expiration dates and acknowl-edged that best evidence to support this resides in The concern that expiration dating may markedly the shelf life extension program (SLEP). Smaller underestimate the actual shelf life of drug studies have addressed the long-term stability of products has been an issue.1–3 The American drug products5,6 and drug substances.7 One study Medical Association (AMA) recently reviewed the determined that four products, captopril tablets, procedures for setting pharmaceutical expiration flucloxacillin capsules, cefoxitin injection, and dates and the clinical and fiscal consequences of theophylline tablets stored under ambient tem- setting such dates.4 The AMA concluded that the peratures maintained at least 98% of label claim actual shelf lives of some products are greater for drug content for 18–170 months past thelabeled expiration dates.5 In a modification to the Correspondence to: Robbe C. Lyon (Telephone: 301-796- 0019; Fax: 301-796-9816; E-mail: robbe.lyon@fda.hhs.gov) industry, the Food and Drug Administration Journal of Pharmaceutical Sciences, Vol. 95, 1549–1560 (2006)ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 (USP) were urged to determine the benefits and valuable source of long-term stability data for a risks associated with lengthening expiration dates and to subsequently conduct longer stability This report summarizes data for 3005 lots testing. In response, all of the data from the SLEP representing 122 drug products generated by the was reviewed and analyzed. As a retrospective SLEP since 1986. Based on stability assessment, analysis, this report summarizes extended drug 88% of the lots were extended beyond their original product stability data collected by the SLEP over expiration date. Of the 2652 lots extended, only 18% were eventually terminated due to failure.
The International Conference on Harmoniza- The rest of the lots are either still active (35%) or tion (ICH) defines shelf life as ‘‘the time period were abated (47%) by the military. The shelf life which a drug product is expected to remain within extension results were summarized in tabular the approved shelf life specification, provided that form. The products were arranged into groups it is stored under conditions defined on the container label.’’8 It is expected that the actualshelf life will slightly exceed the projected labeledshelf life. Pharmaceutical manufacturers are required to assign an expiration date to each drugproduct marketed in the United States. The labeled shelf life is estimated using appropriatestability testing (21 CFR 211.137 and 211.166) The SLEP is a key component of the Medical under current good manufacturing practices as Readiness Strategic Plan as developed by the established and monitored by the FDA. The DOD Health Affairs and the Military Medical stability assessments of both the new drug sub- Departments. The DOD Defense Medical Stan- stance and the new drug product stored under dardization Board (DMSB) oversees the SLEP controlled conditions are required documentation program and acts as an interface between the for submission to the FDA in a new drug applica- military services and the FDA. Pharmaceutical tion (NDA). The assessment follows scientifically drug products sealed in original container clo- based technical procedures as described in the ICH sures are stored under controlled conditions by Q1A(R2) Guidance.8 The initial expiration date is the military services. Certain lots of drug product based on the amount of real-time stability data that are approaching their labeled expiration date (generally from pilot scale batches) for the drug are selected by the DMSB for participation in the product available at the time of approval of the SLEP program. Representative sealed containers NDA. This initial date may later be extended of drug product from a given lot are submitted to contingent upon the receipt of acceptable support- the FDA SLEP coordinator in the Office of ing data from the manufacturer based on acceler- Regulatory Affairs for testing by the FDA field ated stability studies and actual real-time stability labs. The test results are transmitted through the data collected from the first three production FDA SLEP coordinator to the FDA SLEP chemist batches. Drug products marketed in the United in the Division of Product Quality Research States generally have a labeled shelf life of 12– Research (CDER) for analysis. The FDA SLEP The SLEP is administered by the FDA for the chemist evaluates the test results, approves or U.S. DOD9 and recently for the Strategic National rejects the extension of the shelf life for that lot Stockpile (SNS). To maintain a state of readiness, and archives the data, evaluations, and approvals the military maintains, under controlled condi- for the program. Approval for a new expiration tions, large stockpiles of pharmaceuticals sealed in date for that lot of drug product is transmitted their original container closures. A system of through the FDA SLEP coordinator to represen- extending the functional shelf life of these drug products beyond their original expiration date wasinitiated to reduce the high cost of replacing these stockpiles. Based on a comprehensive testingprogram, the shelf life of several drug products Sealed containers of drug product from a given lot has been extended on a lot-by-lot basis. This are sent to an FDA field lab. Samples are program has resulted in substantial savings to subjected to a battery of tests prescribed by the military. In return the FDA has access to a JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 specifications are based on compendial (USP) autoinjectors, the attributes were potency (assay), product release tests or the product release tests degradants, pH, preservatives, injection mecha- as described in the original FDA submission (NDA). If a lot fails any specification based onthe battery of tests, then the shelf life for that lotis expired.
The 122 drug products evaluated by this study were categorized into five groups (see Tab. 1) Extending the shelf life is based on developing a based on shelf life extension data (relative history of real-time stability data for each lot of number of lots initially extended and number of drug product. This data is compiled by continual extended lots terminated). For the products testing within the SLEP. Test results from each assigned to Groups 1 and 2, all lots were extended retest project form a set of real-time data. That beyond their original expiration date. The pro- real-time data is evaluated and an updated ducts assigned to Groups 3, 4, and 5 had some lots expiration period for that lot of drug product that were denied initial extension. Overall, 2650 is predicted using regression analysis. Each (88%) of the 3005 lots were extended past retest attribute generates a particular remaining their original expiration date for an average of expiration period. If each remaining expiration 66 months. Of these, 934 lots (35%) are still period predicted is longer than 1 year, the lot of active and were granted an average extension of drug product is granted a new expiration date. An 62 months, 1237 lots (47%) were abated before ‘‘initial extension failure’’ indicates that the lot failure (dormant) after an average extension of could not be extended past original expiration 70 months, and 479 lots (18%) were terminated date. A ‘‘termination failure’’ indicates that a due to failure after an average extension of previously extended lot eventually fails upon re- 65 months. Of the 479 lots that eventually failed, testing and based on regression analysis predic- no lots failed before 1 year and 312 lots were tions, cannot be extended for an additional year.
further divided into subclasses based on the number of lots tested for each product. For The attributes tested for a drug product varied these products, none of the extended lots were depending on the dosage form. For solid oral drug terminated due to failure (periodic retest). Each of products, the attributes were potency (assay), the 15 products assigned to Group 1A (Tab. 2) had impurities, water content, dissolution, and physi- at least 10 lots evaluated. Of the 466 lots tested, cal appearance. For reconstituted dry powders, the 284 lots (primarily 205 lots of ciprofloxacin tablets) attributes were potency, pH, water content, and are still active in the program and available for physical appearance. For injectable solutions, the use. The average extension for these active lots attributes were potency, impurities, pH, preserva- was 52 months. The remaining 182 lots were not tives, and physical appearance (color, particu- further tested or extended and are categorized as lates). For creams and ointments, the attributes dormant. The average extension for these dormant were potency, pH, and physical appearance. For lots was 67 months. The average extension time for Number of Extended Average Extension Time JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 Group 1A: Products With No Failures (10 Lots tested) aSee ‘‘spray’’ dosage form (Group 5).
bSee ‘‘tablet’’ dosage form (Group 3).
cSee ‘‘autoinjector’’ (Group 2) and ‘‘injection-solution’’ (Group 3) dosage forms.
dSee ‘‘granules’’ dosage form (Group 1B).
eSee ‘‘injection-solution’’ dosage form (Group 2).
all of the lots from these 15 products was are now dormant. The average extension for these 58 months, ranging from 12 to 184 months. The dormant lots was 55 months, ranging from 12 to mean extension times and the ranges of extension 114 months. The tablet dosage form of chlorpro- times for the individual products are provided in the table. For six of the products the extension Each of the 20 products assigned to Group 2 times represent a combination of the active and (Tab. 5) had some lots that could not be repeatedly dormant lots. In these cases, the range for the extended, one attribute eventually failing during active lots is annotated in the table. Other dosage periodic retest. Of the 254 lots tested, 41 lots forms of diphenhydramine HCL, doxycycline eventually failed and were terminated, 211 lots are hyclate, morphine sulfate, potassium iodide, and now dormant and 2 lots are currently active. For sodium chloride were assigned to other groups.
the terminated lots, the amount of shelf life The 25 products assigned to Group 1B (Tab. 3) extension before the failure ranged from 12 to had 5–9 lots evaluated. Of the 164 lots tested, 143 103 months with an average of 53 months.
lots are dormant and 21 lots are still active. The Termination was due to a variety of test failures average extension for the dormant lots was as indicated in the table. The average extension for 61 months and for the active lots was 92 months.
the 211 dormant lots was 59 months ranging from The average extension time for all of the lots from 15 to 137 months. The mean extension times and these 25 products was 65 months, ranging from the ranges of extension times indicated in the table 15 to 278 months. Other dosage forms of ampi- represent a combination of the terminated, dor- cillin, cimetidine HCl, ciprofloxacin, potassium mant, and active lots. The average extension time iodide, and povidone-iodine were assigned to other for these combined lots was 58 months, ranging from 12 to 137 months. The range of extension The 23 products assigned to Group 1C (Tab. 4) times for the two active lots is annotated in the had only 3–4 lots evaluated. All the 85 lots tested table. Other dosage forms of ampicillin sodium, JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 Group 1B: Products With No Failures (5–9 Lots Tested) aSee ‘‘injection-solution’’ dosage form (Group 2).
bSee ‘‘injection-solution’’ dosage form (Group 2).
cSee ‘‘tablet’’ dosage form (Group 1A).
dSee ‘‘tablet’’ dosage form (Group 1A).
eSee ‘‘solution’’ dosage form (Group 4).
chlorpromazine HCl, cimetidine, morphine sul- annotated in the table. The reasons for denying fate, and sodium chloride were assigned to other the extension of the 22 lots was due to a variety of test failures as indicated in the table. Other dosage Each of the 13 products assigned to Group 3 forms of atropine sulfate, doxycycline hyclate, and (Tab. 6) had most lots extended initially (50% morphine sulfate were assigned to other groups.
occurrence) and none of the extended lots were Each of the 16 products assigned to Group 4 terminated due to failure (periodic retest). Of the (Tab. 7) had most lots extended initially (50% 278 lots tested, 256 lots were extended and 22 lots occurrence) and some of the extended lots were were denied extension. Of the extended lots, 81 lots terminated due to failure (periodic retest). Of the are now dormant and 175 lots are currently active 1675 lots tested, 1402 lots were extended and 273 (primarily 159 lots of doxycycline hyclate tablets).
lots were denied extension. Extension denials were The average extension for these dormant lots was due to a variety of test failures as indicated in the 74 months and the average extension for these table. Of the extended lots, 431 lots eventually active lots was 30 months. The average extension failed and were terminated, 519 lots now dormant time for all of the lots from these 13 products was and 452 lots are currently active (primarily 119 lots 44 months, ranging from 12 to 216 months. For of atropine sulfate autoinjectors and 188 lots of three products, the extension times represent a pralidoxime chloride autoinjectors). For the termi- combination of the active and dormant lots. In nated lots, the amount of shelf life extension before these cases, the range for the active lots is the failure ranged from 12 to 266 months with an JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 Group 1C: Products With No Failures (3–4 Lots Tested) aSee ‘‘tablet’’ dosage form (Group 2).
average of 67 months. Termination was due to a failures as indicated in the table. The average variety of test failures as indicated in the table. The extension for the 16 dormant lots was 33 months.
range of extension times for the active lots is The average extension time for all of the lots annotated in the table. The average extension for (terminated and dormant) from these 10 products these dormant lots was 81 months and the average was 38 months, ranging from 14 to 94 months. For extension for these active lots was 79 months. The the individual products, the mean extension times average extension time for all of the lots (termi- and the ranges of extension times in the table nated, dormant, and active) from these 16 products represent a combination of the terminated and was 76 months, ranging from 12 to 266 months. The dormant lots. The spray dosage form of diphenhy- ointment dosage form of povidone-iodine was dramine HCl was assigned to Group 1A.
Each of the 10 products assigned to Group 5 (Tab. 8) had most of the lots denied initial extension (50% occurrence) and some of theextended lots were terminated due to failure (periodic retest). Of the 83 lots tested, 23 lots wereextended and 60 lots were denied extension.
Each product tested was assigned to one of five Extension denials were due to a variety of test groups (as listed in Tab. 1) based on an explicit failures as indicated in the table. Of the extended classification system. Products with no failures lots, 7 lots eventually failed and were terminated, (initial extension failure or termination failure) 16 lots are now dormant, and none of the lots are were assigned to Group 1. This does not imply currently active. For the terminated lots, the that these products will be stable indefinitely.
amount of shelf life extension before the failure Based on past performance, these products may ranged from 17 to 94 months with an average of be considered the best candidates for shelf life 49 months. Termination was due to a variety of test extension. Other factors, including the number of JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 * See See See Active Ac Ac 4-Ep Particula Bactra JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 lots tested and the minimum extension time tested. The combination of lots from atropine surpassed by all lots tested, need to be considered sulfate autoinjectors (687 tested), pralidoxime in evaluating the performance of these products.
chloride autoinjectors (412 tested), and pyridos- Group 1 was divided into 3 subgroups based on tigmine bromide tablets (152 tested) represents the number of lots tested. A greater confidence in 42% of the total lots summarized in this report.
the ability to extend the shelf life was associated Despite the failures, this group was relatively with Group 1A (10 lots tested). The most prolific successful, considering 1404 (84%) of the 1675 lots performer from this subgroup was ciprofloxacin tested were initially extended. Diazepam auto- tablets (242 lots tested with 205 lots still active).
injectors (67 lots tested) had only one lot that failed Two other top performers from this group were initial extension and pralidoxime chloride powder naloxone HCl (all 10 lots extended at least 5 years) (80 lots tested) had only two lots that failed initial and halothane (all 12 lots extended over 4 years).
extension. Of the extended pralidoxime chloride The top performer in Group 1B was potassium powder lots, only two were terminated, but these iodide granules with all five lots still active nearly two lots were extended for more than 8 years.
20 years after the original expiration date.
Other products were successfully extended includ- In addition, all of the lots of calcium chloride ing pralidoxime chloride autoinjectors (97% of lots injection-solution and fentanyl citrate injection- tested), lactated Ringers injection solution (95%, solution were extended more than 5 years. All of 59 lots tested), and pyridostigmine bromide tablets the lots of two products from Group 1C (bupiva- (93% of lots tested). The product exhibiting the caine HCl injection-solution and penicillin G greatest lot-to-lot variability was atropine sulfate benzathine suspension) were extended more than autoinjectors. Hundred and ninety two lots (28%) failed initial extension. The range of extension Although Group 2 had some lots that encoun- times for lots of atropine sulfate autoinjectors tered termination failure, some of these products exhibiting termination failures was 1–11 years.
could be considered quite reliable. Examples from For each of the 10 products assigned to Group 5, this group indicate that if products are stored long most of the lots could not be extended. Lot-to-lot enough, failures are bound to occur. Of the products variability was also evident with this group. Five of with 10 lots tested, notable examples are listed.
the seven penicillin G procaine powder lots could All of the cephapirin sodium powder lots (13 tested) not be extended, while the other 2 lots were were extended at least 4 years with only one extended for more than 5 years. Fourteen of 21 termination, occurring at 8 years and all of the mefloquine HCl tablets lots could not be extended, pancuronium bromide injection-solution lots (13 1 extended lot was terminated at 17 months, while tested) were extended at least 4 years with only one 1 lot was extended for more than 7 years. Of the 83 termination, occurring at 7 years. In addition, lots tested, only 28% could be initially extended.
sulfadiazine silver cream (37 lots tested) had only All of the products in this group were considered one termination, occurring after 4 years and poor candidates and were discontinued from the dextrose injection-solution (23 lots tested) had four terminations, but all occurred after 8 years.
termination failures, but for each product one tofour lots failed initial extension, exemplifying lot- As more lots were tested, the chances that a lot to-lot variability. In some cases the extension would fail increased. This was reflected in the times were quite long. Although two lots of group classifications. Of the 27 products with 15 or cefazolin sulfate powder failed initial extension, more lots tested, 13 products are assigned to Group the eight remaining lots were extended more than 4 compared to only 5 products assigned to Group 5 years. One lot of spectinomycin sodium suspen- 1A. Sulfadiazine silver, now assigned to Group 2, sion failed initial extension, but the seven remain- did not have any failures during the first 11 years ing lots were extended more than 4 years.
on the program. Up until the termination failure of The products assigned to Group 4 had some lots the 27th lot tested, this product was classified as exhibiting initial extension failure and some lots Group 1A. Diazepam autoinjectors, now assigned exhibiting termination failures. The lot-to-lot to Group 4, did not have an initial extension failure variability and the need for systematic testing during the first 6 years on the program. Up until were most evident with this group. This group the initial extension failure of the 48th lot tested, includes products that have been extensively this product was classified as Group 2. Doxycycline JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006 tablets, now assigned to Group 3 did not have an study could be used to select promising potential initial extension failure during the first 6 years on candidates for an extension program.
the program. Up until the initial extension failureof the 13th lot, this product was classified as Group1A. This illustrates that the group classifications may be dependent on the number of lots tested.
This also supports the subclassification of Group The authors are grateful to Christopher D. Ellison of the Division of Product Quality Research foreditorial assistance.
The SLEP has been successful in helping to maintain drug reserves and reducing costs forthe US Military. It is expected that similar 1. Woods M. Drugs may outlast label date. Post-Gazette programs will be implemented to preserve regio- National Bureau. May 30, 2005. Available at: http://www.post-gazette.com/pg/05150/512789.stm.
nal and national pharmaceutical stockpiles. One 2. Cohen LP. Many medicines prove potent for years example is the SNS, formerly the National past their expiration dates. The Wall Street Pharmaceutical Stockpile, managed jointly by the Department of Homeland Security and the 3. Garamone J. Program extends drug shelf-life.
Department of Health and Human Services.10 American Forces Press Service. March 29, 2000 Available at: http://www.defenselink.mil/news/Mar material stock is rotated and kept within potency shelf-life limits. The FDA has issued a shelf life 4. American Medical Association. ‘‘Pharmaceutical extension guidance for federal agencies and state Expiration Dates’’. Report 1 of the Council on Scien- and local governments. A specific guidance was tific Affairs (A-01). July 25, 2001. Available at: issued by CDER presenting FDA’s recommenda- http://www.ama-assn.org/meetings/public/annual01/csa_reports.pdf#search ¼ ‘Pharmaceutical%20Expi- tions on testing to extend the shelf life of stock- piled potassium iodide.11 This guidance describes 5. Stark G, Fawcett JP, Tucker IG. 1997. A study of the how to identify laboratories suitable to conduct stability of some commercial drug dosage forms the tests, how to notify holders of stockpiled drug beyond their expiration dates. Pharm J 258:637–640.
products and end users about changes in expira- 6. Regenthal R, Stefanovic D, Albert T, Trauer H, tion date, and how to distinguish stockpiled Wolf T. 2002. The pharmacologic stability of 35- batches with different expiration dates.
year old theophylline. Hum Exp Toxicol 21:343–346.
7. Scholtissek C, Webster RG. 1998. Long-term stability of the anti-influenza A compounds—amantadine and rimantadine. Antiviral Res 38: The SLEP data supports the assertion that many drug products can be extended past the original 8. International Conference on Harmonization: ‘‘Gui- expiration date, but this additional stability dance for Industry Q1A(R2) Stability Testing of period is highly variable. Due to the lot-to-lot New Drug Substances and Products.’’ Available at: variability, the stability and quality of extended http://www.fda.gov/cder/guidance/5635fnl.pdf.
drug products can only be assured by periodic 9. DOD-FDA Shelf Life Extension Program website.
Available at: http://www.usamma.army.mil/html/ testing and systematic evaluation of each lot. The results of this stability program can only be 10. Strategic National Stockpile website. Available at: related to products that have been carefully stored in their original sealed container closures.
11. FDA: ‘‘Guidance for Federal Agencies and State The class groupings indicate past stability perfor- and Local Governments Potassium Iodide Tablets mance and do not guarantee future performance.
Shelf Life Extension.’’ Available at: http://www.
The classification of products presented in this JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 95, NO. 7, JULY 2006

Source: http://www.ofcaems.org/ds-Stability_Profiles.pdf