curtiskleier-finalproject

Curtis K. Kleier Drexel University Department of Chemistry Drexel University, Philadelphia, PA 19104-2876 Submitted on December 4, 2010
 * Cannabinoid - A New Class of Wonder Drugs? - A Historical Perspective of their Discovery and Synthesis**

Medicinal Marijuana has been brought into public light over the decades as an effective treatment of symptoms directly/indirectly related to AIDS, Cancer, Glaucoma, MS along with additional diseases. This paper hopes to accomplish the following:
 * Abstract**


 * Provide an overview of the molecular structures between the active molecules found in natural marijuana and attempt to identify the common chemically active reaction sites as cited in the referenced articles
 * Review a sampling of literature discussing the effectiveness of medications derived from synthetic analogs when compared to their natural counterparts in treating symptoms related to cancers and the treatment of pain via the CB1 and CB2 receptors[2]

Specifically the comparison of the activity mechanism between the naturally occurring cannabinoids found in marijuana with their synthetic brethren using 1’,1’Dimethylheptyl-∆8-tetrahydocannabinol-11-oic acid (CT-3)[3] as an example compound will be discussed in this review. In addition to this historical comparison a brief interlude of the social issues of legalized marijuana will be covered.

There are a number of political hot topics that surface every election – alcohol, tobacco, national speed limits, i mmigration but there is one which affects us every day. As we gain knowledge of the human body we gain increasing dependence on medications to ease ourincreasing list of aliments. According to the World Health Organization (WHO) of the top 10 causes of death, cancer makes up 10% of all deaths for high-income countries.[4] Commonly accepted treatment procedures for many of these cancers include some form of chemotherapy and radiation whose side effects are often mitigated with additional medications, some which fall under the Federal “illicit” category. One of these medications enjoy an existence of quasi-legality namely, medical marijuana. In the past self-medication using natural marijuana has found some market competition in recent history with the introduction of a synthetic alternative under the brand Marinol.
 * Introduction:**

Marinol (Dronabinol) has been identified as structurally identical to ∆9 –THC, the active component in marijuana as seen at right. This synthesis may have seen the shelves in recent history however the pathway to synthetic cannabinol has been known as early as the 1940’s[5] where these synthetic derivatives utilize the same modes of activity without any of the legal complications. One such synthetic alternative is the molecule CT-3 that maintains the core structure but differs in the functional groups. [6] This core structure has beenaccepted as a dibenzopyrone derived out of the seminal work of J.I Thorton and G.R. Nakamura and defines the pathway towards cannabinol synthesis. [7] The discovery of the potential synthesis mechanism provided the impetus to isolate each compound and determine its activity.

The characterization of the individual compounds found in marijuana provided evidence for the role of the CB-1 and CB-2 receptors in pain management vs. psychotropic effects. The article Antihyperalesic properties by Dyson, et al. indicates that the CB1 and CB2 receptors are responsible for neuropathic and inflammatory pain.[2] Furthermore CB agonist such as THC will act on the pain receptors as a secondary mode of action leaving a desire to create a CB specific agonist without the psychotropic effects of THC. The development of the compound 1’,1’Dimethylheptyl-∆8-tetrahydocannabinol-11-oic acid (CT-3) as an effective treatment for chronic pain has been explored since the early 2000’s.[6] This synthetic alternative derived out the of Burstein’s work[8] which observed current work using THC as a therapeutic drug and attempting to develop a clear alternative.

In order to understand the Cannabinol family it is useful to understand the various structures that are found in marijuana. The first works using marijuana attempted to isolate the individual compounds via vacuum distillation. These individual compounds can be broken down into three families: cannabinols, cannabidiols and tetrahydrocannabinols.
 * Cannabinol Structures**

The first formal compound which gave the family its name was completed in 1896 by British chemists Wood, Spivey and Easterfield.[7] Later on the same group purified the cannabinoid into a crystalline acetate. This characterization process continued on for several decades with the development of several new compounds. One of these groups was a cannabinolactone which can undergo a series of different reactions leading to the partial series seen below. [7]
 * Cannabinol**



This cannabinolactone series accounts for just half of the entire structure of the original cannabinol. In his work, Cahn notes that this core structure must bear other functional groups and proposed the following core structure without being able to prove or synthesize the molecule.[9]



It was not until 1940 when this structure was able to fully synthesized and a clear mechanism was proposed by Adams and Baker using a condensation reaction, dehydrogenation, what appears to be a catalytic cleavage.[5]



This synthesis method allowed Adams to continue work on deriving additional isomers of the cannabinol in which he named cannabidiol.[7] This new compound contained an olivetol which from earlier work helped form the core structure. However it was important to discover the specific pathway as this could lead to the metabolic pathways. Adams proposed the following structures for this cannabinol.

vs. the previous proposed structure of

After comparative analysis he was able to derive that the structure on the left was the correct form. This pathway led him to discover several other analogs. The structures were analogous to cannabinol but did not reflect the same properties as cannabinol. Adams took the opportunity to use the accepted synthesis pathway to synthesis two additional molecules. Adams took a set of four cannabinol-like compounds that were synthesized in the lab and compared them to natural cannabinol.



As seen the compounds are all resorcinol derivatives [10] with varying placement of the carbon chain. In later papers this was confirmed using spectra analysis as being very similar to cannabinol as seen in the following figure.

[10]It was through this spectral analysis that the final structure of the cannabinol was derived.

This sibling of cannabinol is really the older child for the attention is drawn towards the more commonly known cannabinol even though the discovery predated the identification of cannabinol and leads to specific cleavage that may be more indicative of cannabinol’s biological roots. The compound was characterized and isolated using hydrolysis of the marijuana resin.[11] The structure is part of the future cannabinol family as it is a resorcinol derivative. This optically active dihydric phenol compound created two unique c rystalline derivatives along with a dimethyl ether.[12]
 * Cannabidiol **

These compound derivatives share a remarkable similarity to cannabinol and thus to cannabinol which is confirmed via a spectral analysis. The difference is that the cannabinoid spectra are shifted towards the longer wave lengths.[12] The believed cause for the shifting was attributed to the double bonding and the stability due to conjugation before NMR was capable of identifying the specific location of the double bond and thus the overall structure.[13] Previously Adams had already dismissed this location for the double bond due to the fact that it was not supported by the UV spectra. It was through this dismissal that allowed Chvo and Mechoulam to whittle down the choices for the double bond and confirm the orientation through the NMR spectra.

The characterization and synthesis of cannabidiol led to the development of optically active synthetic product which although identical to the natural material, had a biological activity when metabolized to tetrahydrocannabinol (THC) which was 70% more active than the natural compound.[7]

The most famous of the cannabinol family is also the psychotropic active component exists as two natural isomers.[7] The extraction of THC from the plethora of compounds in marijuana was driven by the need to discover a compound that was physiologically active but not toxic as compared to the crude cannabinol.[14] The extraction of the psychotropic compounds was derived from previous extracts resultant in the separation of the other cannabinols. The initial separation of the active compound was significantly hindered and prohibitively time consuming (20% yield) but gave a product that was able to be crystallized.[14] It is important to note that this new product gave similar results to the crude product (as if you started with marijuana) but was over a hundred times more effective and gave results using just .38mg/kg vs. a 20mg/kg dose.
 * Tetrahydrocannabino**l

It should be noted that these were natural derivatives and the activity shown are not representative of synthesized version. However synthesized versions did show similar psychotropic effects in dogs as seen in the work by Todd and expressed by Wollner, et al as measured by the Geyer test for physiological activity.[15] The first synthesis attempts of tetrahydrocannabinol was accomplished by two separate research groups, one led by Adams and the other by Todd.[7] The two groups had similar processes to accomplish the final goal. In the work by Adams and Baker they used the c annabidiol primer and reacted with the traditional methylmagnesium iodide to form the precursor to tetrahydrocannabinol.[16]

Tetrahydrocannabinol synthesized via this intermediary process using methylmagnesium iodide formed products that showed a 1/7th of the activity when compared against natural tetrahydrocannabinol. The work of Desai showed that it was possible to synthesize a molecule where the double bond of the core molecule did not have to be fixed in order for there to be psychotropic activity.[16] The work completed by Wollner showed the specific difference in potency of the various fractions derived from the synthesis of tetrahydrocannabinol using the method proposed by Adam s.[15]



Potency was measured using a common standard measured against a reproducible synthetic analog which is preferred over the natural tetrahydrocannabinol. This paper showed that despite the predisposition of natural marijuana to be composed of single enantiomer vs. the racemic synthesis there is little correlation to the activity of the product. Thus the rotation of the molecule plays little part in the presence of psychotropic ‘marijuana-like’ activity.[17] There exists an entire series of diastereoisomeric compounds that are part of the tetrahydrocannabinol family and have pharmacological activity as seen at left.

It should be noted that the data gained by Adams was initially not on crystalline forms. It was not until the work of Kort and Sieper which isolated the two crystalline products. [7] Additionally the work of Wollner in 1942 and Todd in 1946 does show the levorotary forms does have increased activity in comparison to that of the dextrorotary forms.[18] Conversely racemic tetrahydrocannabinol shows reduced activity.

The formation of racemic tetrahydrocannabinol is a result of the unstable nature of the crystalline cannabidiol precursor which isomerizes in acidic solutions. This reaction pathway from cannabidiol to cannabinol to the final tetrahydrocannabinol is shown below.[19]



This work led the way to controlling the product composition via an easily adjusted mechanism. This directly correlates to being able to control the physiological action while reducing the potency of the marijuana effect.[19] The basis for the pharmacological activity tests lies in the work by Adams and Loewe who in 1949 identified and charted the potency of various homologs of tetrahydrocannabinol. This was important to characterize the increasing number of compounds without any common standards. These increasing number of compounds was the result of Gringard reactions and thus the following table is one of the first attempts to organize potency of tetrahydrocannabinol.[20]



In the early 60’s and 70’s there was a push to discover the psychotropic compound of marijuana which lead to the discovery of ∆9-tetrahydrocannabinol as the active compound. The resultant medical research found that this compound was an effective treatment of a variety of symptoms. As the pharmaceutical companies designed effective routes to synthesis the market began to see specific medications such as Cesamet which addressed nausea and vomiting as a result of cancer treatments. This synthetic direct replacement for ∆9-THC is interacts with the CB-1 receptor in the neural tissues.[21] According to the full prescribing information the effects of the medication can last as long as 72 hours with an average does of no more than 4mg/day. Where Cesamet seems to relieve pain and discomfort via the CB-1 receptor the medication Marinol increases appetite through a unpublished receptor.[22] However this class of cannabinoid compounds has been identified as acting through a series of GPCR namely CB1 and CB2.[23]
 * Cannabinol Pain Receptors and Medical Uses**

These receptors were cloned in the early 90’s by Matsuda (CB1, 1990) and Munro (CB2, 1993) and it was found that mammalian tissues activate these receptors. It was then identified that agonists and antagonists could be developed to interact with these receptors and more importantly the synthetic THC compounds being produced interacted with these receptors.[24] The receptors are located in different systems with CB1 receptors located in central neurons and CB2 located on immune cells. However this association is not exclusive with CB1 and CB2 receptors being expressed in opposite areas. However the specific role of the CB2 receptors located in neuronal cells is not known.[23] This entire system of receptors and signaling agents is referred to the endocannabinoid system.

The current belief is that the CB1 and CB2 receptors release neurotransmitters that are responsible reducing excessive neural activity. The CB1 receptors are located at the nodes connecting the CNS with the peripheral neurons and inhibit the release of transmitters responsible for flight/fight mechanisms.[25] The mechanism for the release of these CB1/CB2 antagonists is triggered via varying levels of calcium levels which causes biological synthesis of the respective antagonist by inhibiting the release of the respective excitatory neurotransmitters. The mode of inhibition for the antagonist is very different for the two CB receptors. CB1 receptors are normally inhibited by the antagonist such as synthetic THC which limits the release of neurotransmitters. On the other hand, CB2 receptors have a tendency of being inhibited because the antagonist is acting on the chemical messengers rather than the neural transmitter via the immune cell migration.

It is this mode of inhibition that becomes of value when comparing synthetic THC with natural marijuana. The molecule ∆9-THC has the ability to act upon both receptors with a certain affinity range dependent on the isomer. The interaction of natural THC with the CB1 and CB2 receptors are lower than the interaction between the synthetic counterparts. This supports the pharmacological activity difference seen earlier where synthetic THC shows a decrease in marijuana-like symptoms. This has been linked to the activation of the CB1 receptor site rather than the CB2 site. This affinity for the CB1 sites by natural THC causes what is now called the tetrad of effects.[23]

This tetrad of effects is indicated by suppression of motor activity, hypothermia, and immobility when exposed to stimuli. This CB1 affinity was confirmed by comparison to known activity using a CB1 specific receptor and charting the tetrad effect. Once the CB1 receptor compatibility is removed the complete expression of the tetrad is not seen but the pain relieving aspects are still present. These receptors are also responsible for the neurotransmitters related to appetite stimulation and management of glaucoma. In recent history there is clinical support that these cannabinoid receptor agonists are effective treatments for postoperative pain and the reduction of spasms as a result of multiple sclerosis and spinal cord injury.

CB1 receptor agonists can also stop the release of other neurotransmitters which cause a CB1 receptor mediated effect. [25] However one must not believe that THC is selective when administered in vivo but rather there is a correlation between the CB1 inhibition and the increase release of neurotransmitters in other neurons. It has been suggested that the evidence shows that there is a CB1 mediated increases in connected neurons at the opposite end of the chain from where there is CB1 inhibition.[24] Additionally this CB1 mediation has been linked to the release of dopamine which is responsible for the reduced reward behavior in rats as it seems this reward behavior is CB1 mediated. This means that agonists such as THC will interact with the CB1 receptor and causes the pleasure centers to be activated, food to taste better and as a result one eats more.

The interaction of THC on the CNS is two-fold, causing both stimulatory and depressive results. This action is seen when THC is used to reduce spasms in some forms of epilepsy but it in other investigations it can cause an increase in spasms. It is suggested that this behavior is a direct result of THC’s ability to both block and activate CB1 receptors and in doing so will reinforce or depress CB1 neurotransmitters. This is a direct result of the low affinity that THC has with the CB receptors and thus is influenced by the density of the receptors at a particular site. This density difference of the receptors has been shown to be true in rat brains and thus CB1 reception and interaction with the G proteins is significantly high in one area then in the other. This affinity difference between the CB1 receptor and the G proteins will decrease the ability of a CB1 agonist like THC to interact with the receptor site. The evidence of these difference is shown by the various symptoms that are expressed when the receptor sites are exposed to THC in various animals. In rats and mice we see an increased motor dysfunction as compared to humans due to the decreased CB1 receptor density in the cerebellum as opposed to the cerebral cortex. Additionally these differences can be seen in rats and mice where application of a CB1 agonist will have an inhibitory effect in one species and a stimulatory effect in the other.

**Future Goals** There are clear clinical implications of this dual nature of partial agonism of the CB1 and CB2 receptors as the downregulation of the CB receptor produces the opposite effect of upregulation in normal situations. However there are specific diseases where this does not happen and the CB1 and CB2 receptors provide a pathway for selective relief. In some animal models we see that this behavior is isolated to specific areas such as in cancerous cells when the CB1 and CB2 receptor expression increases in comparison to the tissue adjacent. However the same increased expression is seen during strokes, epilepsy, spinal injury, ALS, and MS. This selective upregulation of CB1 and CB2 receptors has implications of CB1 and CB2 agonist which may be able to act as therapeutic agents if synthetic compounds can be developed which are specific high potency agonists as compared to natural tetrahydrocannabinol. Any doubt that tetrahydrocannabinol is an effective and potentially valuable medication has been alleviated. Current research must be focused on the control mechanisms of the dual nature of the CB1 and CB2 receptor agonist behavior and the development of additional compounds which express CB1 activity without the psychotropic effects of marijuana. Once such compound, CT-3 has shown promise in the field by reducing pain symptoms without cannabinoid effects. In tetrad testing it was shown that it penetrates the CNS and interacts with neural CB receptors but that the main activity is due to interaction on peripheral CB receptors.

**Social Repercussions** When looking at a medication whose nature is illegal, one struggles to determine the cost benefit of developing the product. As marijuana is considered to be the major drug of abuse and some reports have it coming in as the third largest industry in the US. As a society we seem to have accepted the introduction of Marinol and Cesamet as synthetic alternatives but yet why does there seem to be this huge pushback for reclassifying the drugs. One reason may be the social status of marijuana – in some states the possession of the drug is no more than a misdemeanor rather than a felony whereas drugs such as cocaine and heroin are considered to be felony charges at significantly smaller quantities. Thus looking at the history of cannabinoids the legal ramifications seem to be easing and the future of one day developing pain medications utilizing these CB receptors is upon us. Although there are 15 states that have legalized medicinal use of marijuana science and biochemistry is on the brink of developing additional synthetic alternatives which may make this a moot point from a medical standpoint. History has shown us that chemistry is capable of developing significantly superior products with targeted results more efficiently than nature however this is only possible by using and understanding the facts that nature gives us.


 * 1) ChemSpider, Tetrahydrocannabinol (15266 ChemSpider ID) [Accessed via http://www.chemspider.com/Chemical-Structure.15266.html]
 * 2) Dyson, A., et al., Antihyperalgesic properties of the cannabinoid CT-3 in chronic neuropathic and inflammatory pain states in the rat. Pain, 2005. 116(1-2): p. 129-137. http://dx.doi.org/10.1016/j.pain.2005.03.037 [Accessed via http://www.sciencedirect.com/science/article/B6T0K-4G9Y5BT-6/2/12abec8006c55e2d607fbc21d1c6f94a]
 * 3) Karst, M., et al., Analgesic Effect of the Synthetic Cannabinoid CT-3 on Chronic Neuropathic Pain: A Randomized Controlled Trial. JAMA, 2003. 290(13): p. 1757-1762. http://dx.doi.org/10.1001/jama.290.13.1757 [Accessed via http://jama.ama-assn.org/cgi/content/abstract/290/13/1757]
 * 4) Organization, W.H., The 10 Leading Causes of Death by Broad Income Group. 2004. [accessed via:http://www.who.int/entity/mediacentre/factsheets/fs310_2008.pdf]
 * 5) Adams, R. and B.R. Baker, Structure of Cannabinol. V. A Second Method of Synthesis of Cannabinol1. Journal of the American Chemical Society, 1940. 62(9): p. 2401-2401. http://dx.doi.org/10.1021/ja01866a039 [Accessed via http://dx.doi.org/10.1021/ja01866a039
 * 6) Dajani EZ, L.K., Taylor J, Dajani NE, Shahwan TG, Neeleman SD, Taylor MS, Dayton MT, Mir GN., 1′,1′-Dimethylheptyl-Δ-8-tetrahydrocannabinol-11-oic Acid: A Novel, Orally Effective Cannabinoid with Analgesic and Anti-inflammatory Properties. Journal of Pharmacology and Experimental Therapeutics, 1999. 291(1): p. 31-38. [Accessed via http://jpet.aspetjournals.org/content/291/1/31.full.pdf+html]
 * 7) Thornton, J.I. and G.R. Nakamura, The Identification of Marijuana. Journal of the Forensic Science Society, 1972. 12(3): p. 461-519. http://dx.doi.org/10.1016/s0015-7368(72)70716-1 [Accessed via http://www.sciencedirect.com/science/article/B8JJ5-4SDG8P8-7/2/de4c55053e41b5adc7c3c297d0123498]
 * 8) Burstein, S.H., et al., Synthetic nonpsychotropic cannabinoids with potent antiinflammatory, analgesic, and leukocyte antiadhesion activities. Journal of Medicinal Chemistry, 1992. 35(17): p. 3135-3141. http://dx.doi.org/10.1021/jm00095a007 [Accessed via http://dx.doi.org/10.1021/jm00095a007]
 * 9) Cahn, R.S., CXXI.-Cannabis Indica resin. Part I. The constitution of nitrocannabinolactone (oxycannabin). Journal of the Chemical Society (Resumed), 1930: p. 986-992. [Accessed via http://dx.doi.org/10.1039/JR9300000986]
 * 10) Adams, R., C.K. Cain, and B.R. Baker, Structure of Cannabinol. II. Synthesis of Two New Isomers, 3-Hydroxy-4-n-amyl- and 3-Hydroxy-2-n-amyl 6,6,9-Trimethyl-6-dibenzopyrans1. Journal of the American Chemical Society, 1940. 62(8): p. 2201-2204. http://dx.doi.org/10.1021/ja01865a082 [Accessed via http://dx.doi.org/10.1021/ja01865a082]
 * 11) Adams, R., M. Hunt, and J.H. Clark, Structure of Cannabidiol. III. Reduction and Cleavage. Journal of the American Chemical Society, 1940. 62(4): p. 735-737. http://dx.doi.org/10.1021/ja01861a011 [Accessed via http://dx.doi.org/10.1021/ja01861a011]
 * 12) Adams, R., C.K. Cain, and H. Wolff, Structure of Cannabidiol. II. Absorption Spectra Compared with those of Various Dihydric Phenols. Journal of the American Chemical Society, 1940. 62(4): p. 732-734. http://dx.doi.org/10.1021/ja01861a010 [Accessed via http://dx.doi.org/10.1021/ja01861a010]
 * 13) Mechoulam, R. and Y. Shvo, Hashish--I : The structure of Cannabidiol. Tetrahedron, 1963. 19(12): p. 2073-2078. http://dx.doi.org/10.1016/0040-4020(63)85022-x
 * 14) HAAGEN-SMIT, A.J., et al., A PHYSIOLOGICALLY ACTIVE PRINCIPLE FROM CANNABIS SATIVA (MARIHUANA). Science, 1940. 91(2373): p. 602-603. http://dx.doi.org/10.1126/science.91.2373.602 [Accessed via http://www.sciencemag.org/content/91/2373/602.short]
 * 15) Wollner, H.J., et al., Isolation of a Physiologically Active Tetrahydrocannabinol from Cannabis Sativa Resin. Journal of the American Chemical Society, 1942. 64(1): p. 26-29. http://dx.doi.org/10.1021/ja01253a008 [Accessed via http://dx.doi.org/10.1021/ja01253a008]
 * 16) Adams, R. and B.R. Baker, Structure of Cannabidiol. VII. A Method of Synthesis of a Tetrahydrocannabinol which Possesses Marihuana Activity1. Journal of the American Chemical Society, 1940. 62(9): p. 2405-2408. http://dx.doi.org/10.1021/ja01866a041 [Accessed via http://dx.doi.org/10.1021/ja01866a041]
 * 17) Adams, R., et al., CONVERSION OF CANNABIDIOL TO A PRODUCT WITH MARIHUANA ACTIVITY. A TYPE REACTION FOR SYNTHESIS OF ANALOGOUS SUBSTANCES. CONVERSION OF CANNABIDIOL TO CANNABINOL. Journal of the American Chemical Society, 1940. 62(8): p. 2245-2246. http://dx.doi.org/10.1021/ja01865a508 [Accessed via http://dx.doi.org/10.1021/ja01865a508]
 * 18) Todd, A., Hashish. Cellular and Molecular Life Sciences, 1946. 2(2): p. 55-60. http://dx.doi.org/10.1007/bf02163886 [Accessed via http://dx.doi.org/10.1007/BF02163886]
 * 19) Adams, R., et al., Structure of Cannabidiol. XII. Isomerization to Tetrahydrocannabinols1. Journal of the American Chemical Society, 1941. 63(8): p. 2209-2213. http://dx.doi.org/10.1021/ja01853a052 [Accessed via http://dx.doi.org/10.1021/ja01853a052]
 * 20) Adams, R., M. Harfenist, and S. Loewe, New Analogs of Tetrahydrocannabinol. XIX. Journal of the American Chemical Society, 1949. 71(5): p. 1624-1628. http://dx.doi.org/10.1021/ja01173a023 [Accessed via http://dx.doi.org/10.1021/ja01173a023]
 * 21) Inc, M.P., Cesamet Prescribing Information 2010.
 * 22) FDA, Marinol - FDA Drug Label. 2006. [Accessed via http://www.accessdata.fda.gov/drugsatfda_docs/label/2006/018651s025s026lbl.pdf]
 * 23) Pertwee, R.G., The diverse CB1 and CB2 receptor pharmacology of three plant cannabinoids: Δ9-tetrahydrocannabinol, cannabidiol and Δ9-tetrahydrocannabivarin. British Journal of Pharmacology, 2008. 153(2): p. 199-215. http://dx.doi.org/10.1038/sj.bjp.0707442 [Accessed via http://dx.doi.org/10.1038/sj.bjp.0707442
 * 24) http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2219532/pdf/0707442a.pdf]
 * 25) Pertwee, R.G., The central neuropharmcology of psychotropic cannabinoids. Pharmacology & Therapeutics, 1988. 36(2-3): p. 189-261. http://dx.doi.org/10.1016/0163-7258(88)90106-4