PLG Microsphere Size Controls Drug Release Rate through Several Competing Factors Cory Berkland, 1 Kyekyoon (Kevin) Kim, 2 and Daniel W. Pack 1,3 Received January 2, 2003; accepted March 19, 2003 Purpose. Although the rate of drug release from poly(D,L-lactide–co- glycolide) (PLG) microspheres is often modulated by changing fab- rication conditions or materials, the specific factors directly control- ling the release profiles are often unclear. We have fabricated uni- form rhodamine- and piroxicam-containing microspheres, 10 to 100 m in diameter, to better understand how microsphere size controls drug release. Methods. Drug distribution within the microspheres was examined using confocal fluorescence microscopy. The rate of polymer degra- dation was determined as the change in molecular weight, measured by gel permeation chromatography, during in vitro degradation ex- periments. Further, changes in the surface and interior morphology of the particles during in vitro degradation were investigated by scan- ning electron microscopy. Results. Microsphere size greatly affected drug distribution. Small (∼10-m) microspheres showed an essentially uniform drug distribu- tion. Larger (∼100-m) microspheres showed redistribution of drug to specific regions of the microspheres. Rhodamine partitioned to the surface and piroxicam partitioned to the interior of large PLG mi- crospheres. Further, the rate of polymer degradation increased with microsphere size, possibly the result of a more acidic interior caused by increased accumulation of hydrolyzed polymer products in larger particles. Finally, larger microspheres developed a more porous in- terior structure during the drug release. Conclusions. Microsphere size affects drug release not only through changes in diffusion rates but also through secondary effects includ- ing drug distribution in the particle, polymer degradation rate, and microsphere erosion rates. KEY WORDS: controlled release; zero-order release; uniform mi- crospheres; poly(lactide–co-glycolide); piroxicam. INTRODUCTION Poly(D,L-lactide–co-glycolide) (PLG) microspheres have been widely investigated as delivery devices for a variety of therapeutics because they have several important advantages over conventional dosage forms. For example, biodegradable polymer microspheres that can deliver a therapeutic at a con- stant rate over a prolonged time following a single adminis- tration can avoid peak-related side effects, aid patient com- fort and compliance, provide localized drug delivery and high local drug concentrations, and potentially optimize efficacy. More importantly, the rate at which a drug is delivered can impact its efficacy as much as the identity of the drug mol- ecule itself. The rate of drug release from biodegradable poly- mer devices can be controlled by the size of the device (1–5), the degradation kinetics of the polymer (in turn a function of polymer composition, tacticity, molecular weight, etc.) (5–7), and by variation of several of the important parameters used in the formulation process (8–10). Existing devices can re- lease drug over periods from days to months, and the delivery rate can be adequately controlled for many applications. However, there still exists a need for microsphere systems that can easily and reproducibly provide control of delivery rates including “zero-order” and “pulsatile” release kinetics. The size of biodegradable polymer microspheres is well known to be a primary determinant of polymer degradation and drug release rates. PLG degradation occurs by hydrolysis of the ester bonds and can be autocatalyzed by the accumu- lation of acidic degradation products (11). As particle size increases, surface area:volume ratio decreases, which de- creases both buffer penetration and release of degradation products. Thus, larger particles exhibit a more acidic in- trapolymer pH microenvironment (12) and degrade more rapidly (13). Because the mechanism of drug release is typi- cally diffusion through the polymer phase or through aque- ous-filled pores in the polymer matrix, the decrease in surface area:volume ratio with increasing particle diameter translates into a decrease in drug release rate. Furthermore, the size of the particles can impact the kinetics of the fabrication process. For example, in particles formed by solvent extraction, smaller particles are expected to harden faster (because of their larger surface area:volume ratio), which may impact the structure of the polymer matrix and the distribution of drug within the particle. Although a number of studies of the ef- fects of microsphere size on release rates have appeared in recent years (1–5), these various competing effects have been difficult to discern because of the nonuniformity in the size of microspheres prepared by most existing fabrication methods. We have reported a novel methodology for fabrication of highly uniform polymer microparticles (14). With this pro- cess, we have generated monodisperse PLG microspheres from ∼1 to >500 m in diameter; in particular, we reported microspheres with diameters of 5–80 m wherein the diam- eters of 95% of the particles were within 1.0–1.5 m of the average. When small-molecule drug mimics were encapsu- lated in and released from these uniform microspheres, we found that the particle size impacted the release rates as ex- pected, but also the size had a large effect on the shape of the release rate profiles (15). Release profiles (cumulative amount released vs. time) from microspheres smaller than ∼20 m were concave downward, typical of diffusion- controlled release. However, release profiles from micro- spheres larger than ∼40 m were sigmoidal. Such a shape cannot be explained by diffusion alone. We hypothesized that a time-dependent increase in the effective drug diffusivity, caused by degradation of the polymer chains, could account for the sigmoidal shape. In fact, we have shown that a simple model of Fickian diffusion of drug from the particles incor- porating an exponentially increasing diffusivity [D eff (t)  D eff (0)·exp(kt), where k is a constant characterizing the poly- mer degradation rate] provides an accurate fit for release of piroxicam from 10-, 50-, and 100-m diameter PLG micro- spheres (16). However, several other confounding factors 1 Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois 61801. 2 Department of Electrical and Computer Engineering, University of Illinois, Urbana, Illinois 61801. 3 To whom correspondence should be addressed. (e-mail: dpack@ uiuc.edu) Pharmaceutical Research, Vol. 20, No. 7, July 2003 (© 2003) Research Paper 1055 0724-8741/03/0700-1055/0 © 2003 Plenum Publishing Corporation