The Adjuvant Patch: A Universal Dose Sparing Approach for Pandemic and Conventional Vaccines

August 2, 2007
Danielle Smyla

,
Mimi Guebre-Xabier

,
Gregory Glenn

,
Yin-Hwa Lai

,
Jee Loon Look

,
Sarah Frech

,
Barbara Butler

,
Jianmei Yu

,
Chen Xue Zhang

,
Larry Ellingsworth

,
Robert Seid

,
Armond Mayo

,
Vladimir Frolov

,
Mohammad Al-Khalili

Volume 2007 Supplement, Issue 5

In animal studies, we have demonstrated that the dose of an injected H5N1 vaccine candidate can be significantly reduced by using a skin patch containing E. coli heat-labile enterotoxin (LT) applied over the injection site. LT-activated epidermal Langerhans cells migrate to the nearby draining lymph node and enhance the immune response to the injected antigen. A dry patch formulation has been optimized as a dose sparing strategy for pandemic flu and other vaccines. Iomai Corporation has developed a proprietary stabilizing formulation for the patch that allows use and storage at ambient temperature. The patch withstands temperature extremes during shipment, and is suitable for stockpiling.

ABSTRACT

In animal studies, we have demonstrated that the dose of an injected H5N1 vaccine candidate can be significantly reduced by using a skin patch containing E. coli heat-labile enterotoxin (LT) applied over the injection site. LT-activated epidermal Langerhans cells migrate to the nearby draining lymph node and enhance the immune response to the injected antigen. A dry patch formulation has been optimized as a dose sparing strategy for pandemic flu and other vaccines. Iomai Corporation has developed a proprietary stabilizing formulation for the patch that allows use and storage at ambient temperature. The patch withstands temperature extremes during shipment, and is suitable for stockpiling.

In the event of a bird flu pandemic, existing US capacity to produce vaccine falls far short of the projected need.1–3 This shortfall can be overcome by a multipronged approach: 1) improving vaccine potency by using an adjuvant, thereby decreasing the needed dose, 2) expanding manufacturing capacity, and 3) stockpiling to decrease the lead time for the first available doses.

Iomai has developed a dose-sparing strategy based on a universal immunostimulant (IS) patch concept.4 The new approach applies an IS patch, containing E. coli heat labile enterotoxin (LT) as the adjuvant, at the anatomical site of vaccine injection.5–13 The LT-adjuvant patch activates skin dendritic cells (Langerhans cells) that migrate to the draining lymph node where they exert bystander enhancing effects on the immune response to the vaccine injected in the same draining lymph node field.7,9,10

To improve the efficiency of LT delivery to the Langerhans cells in the epidermis, mild disruption of the stratum corneum (SC), the outermost layer of the human skin, is needed prior to the application of the skin patch. For this skin pretreatment, Iomai has developed a disposable skin preparation system (SPS) device to gently abrade the SC to enhance delivery of LT.14 The SPS consists of a strip-pull device with an abrasive surface that is placed on the skin. As the strip is pulled, the SC is gently disrupted. Strip-pulling is simple, painless, and effective for delivery of vaccine, LT, or both into the skin.

Our preclinical studies using LT-IS patches applied over pretreated skin have demonstrated up to 100-fold dose sparing in animals vaccinated with a bird flu recombinant protein vaccine candidate.5

In addition, the LT-IS patch was found capable of augmenting immune responses to different commercial seasonal flu vaccines in preclinical studies. More importantly, the LT-IS patch has been tested in a Phase 1 trial involving a commercial influenza vaccine.15 In this clinical trial, the LT-IS patch was applied to subjects over 60 years of age receiving their annual influenza vaccine. The clinical data demonstrated that an LT-IS patch can enhance immune responses to influenza vaccination in the elderly. Taken together, the results of the preclinical and clinical studies reinforce the idea of a potential universal dose sparing strategy by using an immunostimulant patch containing E. coli LT.

In the preclinical and clinical studies just described, LT was used in a wet patch format to demonstrate proof-of-concept principles. The wet patch format involves pipetting a liquid LT solution onto a gauze pad placed over the injection site, and subsequently covering the wet gauze with a protective adhesive overlay. When compared to a wet patch format, a dry patch format is more appealing in terms of ease of use in a clinical setting and commercial viability. Also, a dry patch product is expected to demonstrate a better stability profile due to the significant reduction of water-mediated degradation processes.

Recently, we have developed a stabilizing formulation matrix to present biological products in such a dry patch format. Using this formulation platform technology, we have applied LT in a dry formulated patch, and have conducted a clinical trial to compare the effect of a dry versus wet LT patch with respect to skin delivery. The clinical results presented in this paper show that the dry LT patch is equal to or better than the wet patch in terms of vaccine delivery.

We also present real-time stability data showing that the dry LT-IS patch is stable at 2–8 °C for at least 22 months. Based on accelerated and thermal cycling stress data, the dry LT-IS patch has an adequate thermostability profile to allow storage and use at room temperature and to withstand extreme temperatures during shipment and distribution.

A pilot-scale manufacturing operation has been custom-built specifically for making LT and other vaccine-based patches. Scale-up work is currently under way to produce and stockpile 150 million LT-IS patches.

Preclinical and Clinical Studies

Dose Sparing for Bird Flu Vaccine Candidate

In a mouse study, the LT-IS patch provided up to 100-fold dose sparing for a pandemic flu vaccine candidate (Figure 1). For this study, three groups of mice were immunized intradermally (ID) with a recombinant hemagglutinin (rHA) protein representing the A/Vietnam/1203/2004 (H5N1) strain at 1.5 μg, 0.15 μg, and 0.015 μg. Half of each group had an LT-IS patch (containing 10 μg LT) placed over the injection site and the other half received a placebo (no LT) patch. Following a two-dose regimen at Day 0 and Day 14, the rHA immune responses were determined two weeks after the second dose. As indicated in Figure 1, the group receiving the 0.015 μg dose with the LT-IS patch had an antibody response (47,491 ELISA units) comparable to the group receiving the 1.5 μg dose ID alone (20,142). This data suggests that a 100-fold reduction in vaccine may be achieved when the vaccine is delivered intradermally with the LT-IS patch. Data collected in the same study suggests that a 10-fold reduction in vaccine can be achieved when the vaccine is delivered by intramuscular injection (IM) (data not shown).

Figure 1. Dose sparing in the mouse model using rH5 (Anti-A/Vietnam/1203/2004). Mice received two intradermal immunizations 14 days apart of 0.015, 0.15, or 1.5 μg rHA and IS patches containing either 10 μg LT or a placebo with no LT. Serum IgG titers were determined by an ELISA method. Results of individual animals (open circles) and the group geomeans (bars) are shown. Groups receiving LT have statistically greater titers than groups receiving placebo patches.

An additional preclinical study was conducted in guinea pigs using the same H5N1 rHA antigen in doses ranging from 0.001–10 μg. Both IM and ID routes were used in a two-dose prime/boost regimen. Animals received an IS patch containing 10 μg of LT or a placebo patch at the time of boost. A 10-fold dose sparing was observed using the LT-IS patch for the IM route (data not shown), and ≥10-fold using the ID route (Figure 2). Taken together, the preclinical data in both animal models show that the LT-IS patch can dose spare 10– to 100–fold.

Figure 2. Dose sparing in the guinea pig model using rH5 (Anti-A/Vietnam/1203/2004). Animals received two intradermal injections 21 days apart of either 0.001, 0.01, 0.1, 1, or 10 μg rHA and an IS patch containing either 10 μg LT or a placebo with no LT at the time of second injection. Serum IgG titers were determined by an ELISA method two weeks after the final immunization. Results of individual animals (5–6 per group, open circles) and the group geomeans (bars) are shown.

Universal Adjuvant Effect for Seasonal Flu Vaccines

We also evaluated the LT-IS patch effect using different manufacturers' influenza vaccines and demonstrated that the adjuvant effect is not antigen-dependent. Four commercial annual vaccines were evaluated: FluShield, Fluzone, Fluvirin, and InFlexal. Trivalent influenza vaccine at 4.5 μg HA dose per virus strain was injected ID into mice with and without an LT-IS patch (10 μg LT). The mice were immunized twice, two weeks apart, with serum collection two weeks after the second immunization. The animals receiving the LT-IS patches had significantly higher anti-influenza IgG titers (as determined by ELISA) than those receiving flu vaccines with placebo patches (no LT). Only anti-A/New Caledonia serum IgG titers are shown in Figure 3 as representative of IgG response in all three strains. Figure 3 indicates that influenza vaccines obtained from different manufacturers were enhanced equivalently with an LT-IS patch and reinforces the idea of a potential universal dose sparing strategy using an LT-IS patch.

Figure 3. IS patch effect of different seasonal flu vaccines in the mouse. Mice received two intradermal immunizations, 14 days apart, of a commercially available seasonal flu vaccine and IS patches containing either 10 μg LT or a placebo with no LT. Serum IgG titers were determined by an ELISA method. Results of individual animals (8 per group, open circles) and the group geomeans (bars) are shown. Groups receiving LT have statistically greater titers than groups receiving placebo patches.

Improved Immune Responses in The Elderly

An LT-IS patch was also tested in humans vaccinated with a commercial influenza vaccine.15 In this clinical study, three groups of approximately 55 volunteers were vaccinated with a standard injected flu vaccine consisting of A/New Caledonia, A/Panama, and B/Shangdong. The first group consisted of healthy young adults; the second and third groups consisted of elderly adults over 65 years of age. In the third group, the elderly subjects were vaccinated with the flu vaccine in combination with the LT-IS patch (50 μg LT). Three weeks following the vaccination, the hemagglutination inhibition (HAI) responses were measured. Figure 4 shows the percentage of seroconversion based on a four-fold rise in HAI titers. The percent change in seroconversion was highest for the healthy young adults (first group), followed by the elderly group (third group) receiving the LT-IS patch. Overall, the LT-IS patch improved the elderly seroconversion rates by 23%, 18%, and 12% percentage points for A/New Caledonia, A/Panama, and B/Shangdong, respectively, as compared to the second group which received the vaccination alone. This study shows that an LT-IS patch applied onto the skin after an IM injection can augment the immune response, as predicted from animal models.

Figure 4. Seroconversion rates of patients in a human clinical trial testing IS patch performance in elderly. Young adults had 69%, 56%, and 61% seroconversion rates for A/New Caledonia, A/Panama and B/Shandong respectively. These rates were significantly higher or showed a trend compared to corresponding seroconversion rates in the elderly without a patch (40% [p=0.005], 36% [p=0.06], and 38% [p=0.03], respectively). The addition of an IS patch to the elderly improved these seroconversion rates to 63%, 54%, and 50% respectively, reaching significance in the A/New Caledonia strain (p=0.01) and a trend for A/Panama (p=0.08). The seroconversion rates in elderly receiving the patch were not significantly different from those in healthy adults, and represent an absolute improvement of 23% (A/New Caledonia), 18% (A/Panama) and 12% (B/Shandong) in seroconversion of the elderly over those not receiving the patch.

Wet Versus Dry Patch Formulations

The preclinical and Phase 1 studies used the LT-IS patch in a wet patch format. A dry patch offers advantages over a wet patch in terms of ease of use in a clinical setting (i.e., fewer manipulative steps required for administration), commercial viability (i.e., production, packaging, and distribution), and stability (i.e., dry product is expected to be more stable than liquid product due to removal of bulk water). Recently, we developed a stable formulation matrix and process technology to compound adjuvants, as well as vaccine antigens of interest, into a dry patch format. Using this formulation platform technology, we have prepared a dry formulated LT patch and have conducted a clinical trial to compare the effect of a wet versus dry LT patch on skin delivery.

For the clinical evaluation, 160 human subjects were given a 50 μg LT dose in either a wet or dry patch. Each subject was immunized twice, at Day 0 and Day 21. Blood serum samples were collected from each subject on Days 7, 14, 21, 28, 35, and 42, for determination of LT IgG and LT IgA titers.

Factorial analysis using LT IgG titers and fold rise (ratio to baseline titers) indicated that the dry patch produced statistically significantly higher titers and fold ratios of LT IgG and IgA than did the wet patch at all timepoints from Day 14 onward (Figure 5). This study demonstrated that the dry LT patch is equal to or better than the wet patch at the same dosage in terms of vaccine delivery. We surmise that the dry patch formulation is readily solubilized by transepidermal water loss, resulting in a greater LT concentration gradient at the surface of the skin compared to the wet patch. This higher concentration gradient provides the thermodynamic driving force to deliver more adjuvant into the skin.

Figure 5. Human subjects received either a "wet" or "dry" patch containing 50μg LT on Day 0 and Day 21. LT IgG titers were measured on Days 7, 14, 21, 28, 35, and 42, and compared to baseline titers. The mean fold rise in LT IgG is indicated by the red and blue lines for the "dry" and "wet" patch, respectively.

Dry Formulation Optimization and Patch Design

Since the clinical trial involving the dry versus wet patches, our laboratory has continued to optimize the dry formulation matrix and to develop the final patch format for transcutaneous delivery. To this end, our current dry patch contains a proprietary stabilizing matrix that is compatible with biological molecules and vaccine antigens. The dry patch is prepared by dosing a small volume (20–40 μL) of the drug substance blend onto an absorbent matrix with a surface area ranging from 1–3 cm2 , followed by drying under moderate conditions in a convection oven. The LT-dosed matrix is assembled with an occlusive backing, release liner and adhesive overlay, and sealed in a nitrogen-purged foil pouch for storage at 2–8 °C (Figure 6). When applied to the skin, the dry patch formulation is readily hydrated by skin moisture (transepidermal water loss) and effectively releases the antigens, adjuvant, or both into the skin. The patch is removed after six hours of wear.

Figure 6. Schematic representation of the IS patch showing patch components and application to the arm.

LT Dry Patch Thermostability

The drug substance(s) and excipients are extracted from the patch and this extract is then used in the various analytical assays. Only key stability-indicating assays will be discussed in this paper.

Real-Time and Accelerated Studies

The nitrogen-purged foil pouches containing the dry LT patches were placed under real-time (2–8 °C) and accelerated conditions (25 °C and 40 °C) in controlled temperature chambers. Figure 7 shows the stability of a representative 50 μg/dose lot manufactured under GMP conditions. The biological potency, as measured by the Y-1 adrenal cell assay,16 showed the LT patch meeting the specification of >3.6 log, μg-1 for all test points, i.e., 22 months for 2–8 °C, six months for 25 °C and one month for 40 °C. The LT content, as measured by SE-HPLC, met the 75% of label claim specification for all test time points for patch products stored at 2–8 °C (up to 22 months) and at 25°C (up to nine months, the endpoint of accelerated testing). At the higher temperature of 40 °C, the LT content met specification after one month of storage; however, LT loss was significant thereafter. The moisture content of this lot throughout the stability studies remained at about 6%. Based on up-to-date stability data, there has been no significant trend of LT patch deterioration at 2–8 °C; thus, the LT patch appears to be a good candidate for stockpiling for a future pandemic situation.

Figure 7. Real time and accelerated stability of a single lot of LT IS patches. Patches were stored at 2–8, 25, and 40 °C and assessed periodically for LT content (by SE-HPLC, top panel) and potency (by Y-1 cell assay, bottom panel).

Temperature Excursion Studies

Thermal cycling studies were performed according to the 1998 FDA draft guideline and 2003 ICH Guidance for Industry for stability testing of packaged drug products17–20 in order to gain insight into the tolerance of dry LT patches to temperature excursions that may be encountered during shipping and distribution. Based on the specific product storage temperature, the product is cycled from low to high temperatures for several days, and the exposure is repeated over three cycles. Since we were interested in refrigerated and possibly room temperature distribution, the dry LT patches were exposed to thermal cycling conditions designed for refrigerated and room temperature stable products as outlined in Figure 8. For the thermal cycling studies, we used 50 μg LT patches (GMP lot) stored at 2–8 °C for approximately 6 months. For each thermal cycle, three patches were used and the average LT content reported.

Figure 8 also shows the results of the thermal cycling study as measured by the SE-HPLC content assay (top panel) and Y-1 cell potency assay (bottom panel). After 12 days of thermal cycling between -20 and 25 ºC and -20 and 40 ºC, the averaged data shows no statistically significant decrease in Y-1 cell potency or LT content in the patches. The results of this study suggest the possible use of expanded shipping conditions for this product (beyond 2–8 ºC).

Figure 8. Thermal cycling study for packaged dry LT patches. Conditions were modeled for products targeting both refrigerated and room temperature storage. The temperature cycling regiment is listed in the table. LT was extracted from patches and quantified by SE-HPLC (top panel) and assessed for potency by Y-1 cell assay (bottom panel). SE-HPLC results are reported as the LT percent of the label claim and are the average of 3 patches. Potency results are reported as log, μg-1. Four control groups held at a constant temperature for the study duration were run. HPLC and Y-1 results from thermal cycled samples were not significantly different from those reported for control samples kept at 2-8 °C (as determined by student t-test, data not shown).

Water Activity of The Dry Patch

Water activity, rather than moisture content, is used commonly by microbiologists and food technologists to assess criteria for safety and quality of food products.21–23 Water activity, aw, measures the free water in a product (i.e., in our case, the dry LT patch); whereas, moisture content measures both free and bound water. According to USP <1112>,24 when the water activity of a product decreases below 0.60, microorganisms cannot obtain the water necessary to support their growth. Most bacterial growth is inhibited below aw~0.91; most yeasts cease growing below aw=0.87, and most molds do not grow below aw=0.80. The aw of the dry IS patch was determined by LabMaster-aw (manufactured by Novasina Instruments) and found to be around 0.35±.03, which is significantly lower than the absolute limit of microbial growth of aw=0.60. Thus, the dry IS patch formulation will not support microbial growth during storage in a nitrogen-purged sealed foil pouch.

Scale Up for Pandemic

A pilot-scale patch production line has been custom built specifically for making LT, as well as other vaccine antigen dry patches. The pilot-scale facility can produce 20,000 units per lot using an automated continuous-based dryer and laminator or packaging equipment. Briefly, the formulated LT blend is dispensed onto circular patch matrixes, which have been previously bonded to an occlusive backing film web roll (Figure 9). The LT-dosed, patch matrix and backing web is continuously fed into a drying oven at a set speed. The resulting dried patch web is then fed into a patch assembly machine and is laminated (sandwiched) between an adhesive overlay and a release liner. The assembled patch is finally inserted in a foil pouch, which is purged and sealed under nitrogen. The final design of the patch product is shown in Figure 6.

Figure 9. Scale-up process for manufacturing 20,000 units per lot using two automated lines.

The patch manufacturing processes described above are inherently scalable to meet pandemic flu needs. Capacity can be increased many-fold by widening the patch film web and dryer. Moreover, by increasing the dryer length, the speed of the patch web could be increased during oven drying. Finally, extra lines of the automated dryers and laminator or packaging equipment can be added to increase output capacity. Scale-up planning is currently under way to identify critical design parameters to produce and stockpile 150 million LT-IS patches within a six month period.

Summary

A dose sparing strategy for vaccines is being developed with an adjuvant (immunostimulant) skin patch. This strategy uses E. coli heat-labile enterotoxin protein (LT) as an adjuvant, and application of the dry LT-IS patch over the vaccine injection site. Preclinical and clinical trials have demonstrated proof-of-concept for the LT-IS patch. Extensive formulation development work has been carried out to identify a stable dry formulation matrix for LT as well as for vaccine antigens. Based on ongoing stability studies, the dry LT-IS patch is stable for 22 months at 2-8° C and up to 6 months at 25 ° C. Moreover, it was shown to be stable against multiple thermal cycles between -20 ° C and 40 ° C (two days per cycle), demonstrating that the LT-IS patch can tolerate temperature excursions during shipping and distribution. We have found that maintaining a low water activity in the patch is critical for ensuring product quality and eliminating microbial growth. The formulation characteristics of the dry LT-IS patch, including its inherent stability, and its use as a dose sparing strategy by which a smaller dose of injected vaccine can produce an effective immune response, make the patch an excellent candidate for mass production and for stockpiling in the event of a flu epidemic. We anticipate that the LT-IS patch will become part of the regular medical arsenal for combating infectious diseases such as seasonal influenza and pandemic flu.

Acknowledgement

The authors wish to thank Ms. Wanda Hardy for assistance with manuscript preparation.

Jee Loon Look, PhD, is the director of formulation and process development at Iomai Corporation, Gaithersburg, MD, 301.556.4500, jlook@iomai.com.

References

1. HHS pandemic influenza plan. US Dept. of Health and Human Services. 2005 Nov. Available from: http://www.hhs.gov/pandemicflu/plan/

2. Webby RJ, Webster RG. Are we ready for pandemic influenza? Science 2003 Nov 28;302(5650):1519–22.

3. Wood, JM. Developing vaccines against pandemic influenza. Philos Trans R Soc Lond. B. Biol. Sci. 2001;356(1416):1953–60.

4. Guébre-Xabier M, Hammond SA, Epperson DE, Yu J, Ellingsworth L, Glenn GM. Immunostimulant patch containing heat labile enterotoxin from E. coli enhances immune responses to injected influenza vaccine through activation of skin dendritic cells. J. Virol. 2003;77(9):5218–5225.

5. Glenn, GM, Frech, SA, Kenney, RT, Ellingsworth, LR. Transcutaneous immunization: antigen and adjuvant delivery to the skin. In: Bronaugh RL, Maibach HI, editors. Percutaneous absorption: drugs, cosmetics, mechanisms, methods, 4th ed. New York: Informa Healthcare; 2005.

6. Glenn GM, Kenney RT. Mass vaccination: solutions in the skin. Curr. Top. Microbiol. Immunol. 2006;304:247–268.

7. Glenn GM, Kenney RT. Transcutaneous immunization. In: Levine MM, Kaper JB, Rappuoli R, Liu MA, Good M, editors. New generation vaccines, 3rd ed. New York: Marcel Dekker, Inc.; 2003.

8. Belyakov IM, Hammond SA, Ahlers JD, Glenn JA, Berzofsky JA. Transcutaneous immunization induces mucosal CTLs and protective immunity by migration of primed skin dendritic cells. J. Clin. Invest. 2004;113:998–1007.

9. Glenn GM, Kenney RT, Ellingsworth LR, Frech SA, Hammond SA, Zoeteweij, JP. Transcutaneous immunization and immunostimulant strategies: capitalizing on the immunocompetence of the skin. Expert Rev. Vaccines 2003;2:253–267.

10. Glenn GM, Kenney RT, Hammond SA, Ellingsworth LR. Transcutaneous immunization and immunostimulant strategies. Immunol. Allergy Clin. North Am. 2003;23:787–813.

11. Hammond SA, Walwender D, Alving CR, Glenn GM. Transcutaneous immunization: T-cell responses and boosting of existing immunity. Vaccine 2001;19:2701–2707.

12. Glenn GM, Taylor DN, Li X, Frankel S, Montemarano A, Alving CR. Transcutaneous immunization: A human vaccine delivery strategy using a patch. Nature Med. 2000;6:1403–1406.

13. Glenn GM, Scharton-Kersten T, Vassell R, Matyas GR Alving CR. Transcutaneous immunization with bacterial ADP-ribosylating exotoxins as antigens and adjuvants. Infect. Immun. 1999;67:1100–1106.

14. Glenn GM, Villar CP, Flyer DC, Bourgeois AL, McKenzie R, Lavker RM, Frech SA. Safety and immunogenicity of an enterotoxigenic Escherichia coli vaccine patch containing heat-labile toxin: use of skin pretreatment to disrupt the stratum corneum. Infect. Immun. 2007;75:2163–2170.

15. Frech SA, Kenney RT, Spyr CA, Lazar H, Viret JF, Herzog C, Glück R, Glenn GM. Improved immune responses to influenza vaccination in the elderly using an immunostimulant patch. Vaccine 2005;23(7):946–50.

16. Sack DA, Sack RB. Test for enterotoxigenic Escherichia coli using Y1 adrenal cells in miniculture. Infect. Immun. 1975;11:334–336.

17. US Food and Drug Administration. Guidance for industry. Stability testing of drug substances and drug products. Rockville, MD; June 1998.

18. Lucas TI, Bishara RH, Seevers RH. A stability program for the distribution of drug products. Pharm. Technol. 2004:68–73.

19. International Conference on Harmonization. Q1D, Bracketing and matrixing designs for stability testing of new drug substances and products. Geneva, Switzerland; Feb. 2002.

20. International Conference on Harmonization Code Q1A(R2). Stability testing of new drug substances and products. Geneva, Switzerland; Feb.2003.

21. Water activity. In: Official methods of analysis of the AOAC. 13th ed. AOAC International 32.004–32.009; 1980.

22. Troller JA, Bernard DT, Scott VN. Measurement of water activity. In: Speck ML, editor. The compendium of methods for the microbiological examination of food. Washington, D.C.: American Public Health Association, 1984. p. 124–134

23. Ayerst G. Water activity-its measurement and significance in biology. Int. Biodeter. Bull. 1965;1:13–26.

24. United States Pharmacopeia. Application of water activity determination to nonsterile pharmaceutical products. USP 2007 <1112>.