Original Author: Paul Perry, Ph.D, BCPP
Latest Revisers: Paul Perry, Ph.D., BCPP, Brian C. Lund, Pharm.D.,
BCPP.
Creation Date: 1996
Last Revision Date: February 2002
Peer Review Status: Internally Peer Reviewed
LITHIUM
Serum Lithium Concentrations
Lithium dosing requirements vary depending upon whether the patient is acutely manic or euthymic. Current data suggests the lithium level required to control an acute manic episode is significantly greater than the level necessary to prevent subsequent episodes of either depression or mania.
Acute Mania
Prien et al (1971) studied the relationship between serum lithium concentrations and clinical response in 116 acutely manic patients treated with lithium for a period of three weeks. They observed that patients with levels exceeding 1.4 mEq/l experienced no greater improvement in their manic symptoms than patients with lower serum lithium concentrations. Although a few of the most ill patients showed either improvement or complete remissions with serum concentrations above 1.4 mEq/l, a similar proportion of very ill patients responded to serum concentrations below 1.4 mEq/l while experiencing considerably fewer ADRs. Additionally, none of the patients with serum lithium concentrations below 0.9 mEq/l experienced a complete remission. Of the mildly active patients with concentrations >0.9 mEq/l the failure and remissions rates were 15% and 40% respectively while for patients with concentrations < 0.9 mEq/l the respective failure and remission rates were 33% and 0%. Thus, it was concluded that a serum concentration 0.9 to 1.4 mEq/l is required for treating acute mania. According to the authors' experience lithium carbonate doses of 150 to 4200 mg/day are required to achieve this level. Depending on the severity of the manic symptoms, control of the attack is usually obtained within four to ten days after the start of drug treatment. Relapse occurs within as few as two days if the lithium is abruptly discontinued during the manic phase.
Stokes et al (1976) examined the effect of three different lithium dosages (low, medium, high) versus placebo in acute manic patients. They treated 32 patients with a low lithium dose (0.5 mEq/kg/d) and serum levels up to 0.4 mEq/l or placebo to an ABAB crossover design. A second group of 36 patients were treated with either a high dose (0.72 mEq/kg/d) and serum levels up to 1.06 mEq/l or with a low lithium dose (0.24 mEq/kg/d) and serum levels not exceeding 0.40 mEq/l also in an ABAB crossover design. Forty-two patients completed the study such that 38 received the low dose, 47 the medium dose, and 40 the high dose. Compared to baseline only the high dose group improved clinically.
Manic patients in the midst of an acute episode require and tolerate higher lithium doses due to an increased lithium clearance. These levels may not be tolerated once the manic episode begins to abate. Therefore, frequent lithium levels, normally drawn twice weekly, are required in the treatment of an acute manic episode because of the possibility of a rapid rise in lithium levels as the episode begins to resolve. Finally, in highly active manics, it is best to combine lithium with an antipsychotic such as haloperidol to control psychomotor behavior until the lithium begins to exert its effect. The antipsychotic should be discontinued as soon after clinical control occurs so to minimize the patient's exposure to discomforting extrapyramidal side effects.
Prophylaxis
The ideal prophylactic serum lithium concentration is a question of considerable debate. Originally, most clinicians recommended a concentration range of 0.8 to 1.2 mEq/l. However, data from both Hullin (1979) and Coppen et al (1983) suggested that lower serum lithium concentrations were as effective. Utilizing a divided daily dosing schedule, Hullin (1979) found that affective illness relapse rates were the same for patients with lithium doses producing lithium concentrations above 0.6 mEq/l as were the relapse rates of patients with concentrations between 0.4 to 0.6 mEq/l. However, concentrations below 0.4 mEq/l were associated with increased relapse rates. Coppen et al (1983) studied the effect of lowering the serum lithium concentration in 72 patients with recurrent affective illness. The patients were randomly allocated to one of three lithium concentration groups, i.e., 0.8 to 1.2 mEq/l, 0.6 to 0.79 mEq/l, or 0.45 to 0.59 mEq/l. Patients were followed for at least one year. Interestingly, the patients who experienced the greatest decrease in affective illness morbidity were the patients with the lowest serum lithium concentration. No change in affective illness morbidity was observed in the group where the lithium dose was not altered. Most recently, Abou-Saleh and Coppen (1989) randomly allocated 90 bipolar and unipolar patients with recurrent affective illness to one of three lithium concentration groups, i.e., > 0.79, 0.6 to 0.79 mEq/l, or 0.45 to 0.59 mEq/l. Patients were followed for one year. There were no differences between the three groups with respect to dose and affective morbidity. Older patients (>59 years) patients experienced a poorer outcome following dosage reduction. The low-level group had significantly fewer tremors, a decrease rather than increase in weight, reduced polyuria, i.e., output, and lower TSH levels. Gelenberg et al (1989) followed 94 bipolar patients whose lithium concentrations were adjusted to a range of 0.8 to 1.0 mEq/l or 0.4 to 0.6 mEq/l. The relapse rate in the high level group was only 13% whereas the rate was 2.6 times higher (38%) in the low dose group. Unlike, the Abou-Saleh study the authors did not take into account the effect age in an attempt to explain the discrepancy. Although controlled for, the authors also found that a greater number of prior episodes, a prior manic episode, and a shorter length of remission before intake contributed to a two- to threefold increase in risk of relapse. However, he did not control for polarity of the index episode. Vestergaard et al (1998) unsuccessfully attempted to replicate Gelenberg's (1989) findings. Seventy-six bipolar and unipolar patients requiring maintenance therapy were randomized to either high-dose (0.8-1.0 mEq/L) or low-dose (0.5-0.8 mEq/L) lithium maintenance therapy and followed for up to two years. For the bipolar patients, the recurrence rates did not differ between the high-dose (7/33) and the low dose (2/24) group. Additionally, there was no difference in the recurrence rates for the unipolar patients.
Additionally, our own intent-to-treat analysis showed a 53% relapse rate in the low dose group and a 32% relapse rate in the standard dose group. This finding was not significant (Fisher's Exact Test, p = 0.06) although the trend is obvious. Overall, these data suggest that serum lithium concentrations of 0.4 to 0.6 mEq/l on a divided daily dose schedule or 0.45 to 0.59 mEq/l on a single daily dose schedule are appropriate in the prophylactic treatment of affectively ill patients although higher concentrations are required in the elderly. A reasonable clinical strategy would be to instruct patients that because they may be at higher risk of relapse at lower concentrations, they should increase their dose by 1.5 times at the first signs of any manic or depressive symptoms and then slowly taper downward to the lower concentration as the symptoms resolve. This is an extremely important bit of patient education since the Gelenberg data was later reanalyzed and found that the occurrence of subsyndromal symptoms of affective illness increased the risk of relapse four-fold regardless of the lithium level (Keller et al 1992).
Serum Lithium Concentration Sampling
The serum lithium concentration recommended for the treatment of acute mania or the prophylactic treatment of affective illness is based on the lithium concentration at steady state drawn 12 hours after the last dose. To understand the clinical significance of this recommendation, one must be cognizant of several lithium pharmacokinetic parameters, i.e., absorption rate, tissue distribution rate, and elimination half-life. The serum lithium concentration versus time curve presented in figure 1 demonstrates all three pharmacokinetic parameters. For any given dose, absorption is represented by the initial ascending portion of the curve; distribution is represented by the terminal ascending and initial descending portion of the curve; and the elimination half-life is represented by the terminal descending portion of the exponential curve. Practically, the most obvious method to determine the ideal time to sample a drug's concentration in the serum is to inspect a curve such as shown in figure 1 and determine the most stable area of the curve, i.e., the area of the curve where the serum concentration has the least flux. Obviously, the least dynamic area of the curve in figure 1 is during the terminal elimination portion. Generally, following absorption, lithium is distributed at a somewhat slower rate such that distribution is complete within six to ten hours (Amdisen 1977). Thus, in the case of lithium, it is obvious serum sampling should not be drawn prior to ten hours after the last dose. Therefore, twelve hours has been selected as the ideal time to draw serum lithium concentrations.
Figure 1. Serum concentration versus time curve for lithium.
Steady-state conditions for lithium require that the patients not have their serum lithium concentration measured until at least four to five half-lives have elapsed. Four half-lives indicates that the patient has reached a serum lithium concentration of 93.75% of its maximum whereas five half-lives implies that the serum concentration is at 96.88% of its maximal attainable concentration for a particular maintenance dose. Thornhill and Field (1982) has estimated the half-life of lithium in 40 euthymic psychiatric patients to range from 15-55 hours. However, we have measured half-lives as short as 8 hours in extremely agitated acutely manic patients who require little or no sleep and thus have significantly higher lithium clearance rates. Assuming an individual patient's lithium half-life is unknown and half-lives range from 15-55 hours for patients with normal renal function, approximately 220 hours, i.e., (55 x 4) hours or 9 days may need to elapse before a true steady state lithium measurement can be obtained. However since the mean of the above population was 30 ± 10 hours, a week is probably a sufficient amount of time to wait in the majority of patients.
Dose Regimen Table 1 demonstrates the effect of a regular-release lithium dosage form on the 12-hour steady-state serum lithium concentration. This table illustrates two points. First, there is a relatively small difference between the 12-hour serum lithium steady-state concentrations for the q8h and q12h schedules but a considerably larger difference between these two schedules and the q24h schedule. Thus, the greater the increase in the dosing steady-state concentrations. Second, as the half-life increases, the percent increase in the 12-hour steady-state concentration decreases. This information is important in clinical situations where a patient on a divided daily dosing schedule is being converted to a single daily dosing schedule. If the clinician assumes a 24 hour half-life for the patient, then an approximate 20% increase in the 12-hour steady-state serum lithium concentration ought to be anticipated. However, as an examination of the data in Table 1 suggests, this figure can range from 12 to 33%.
|
|
|||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Figure 2 presents a computer simulation model of a serum lithium concentration versus time curve for a slow-release lithium product. It is noted that the single daily dosing schedule, although producing a higher 12-hour steady-state concentration, results in a significantly lower serum lithium trough concentration. Because of the clinically significant polyuria caused by lithium in at least 10% of patients, this is an extremely important observation.
Figure 2. Effect of dosing on 12-h and 24-h steady-state serum lithium concentration.
Mellerup et al (1985) have found that urine volume positively correlates only with the trough or minimum serum lithium concentration. Urine volume did correlate with lithium dose, patient age, length of treatment, maximum serum lithium concentration, or 12-hour serum lithium concentration. This finding implies that the only direct change in lithium therapy that can result in a decrease in lithium-induced polyuria is changing patients from a divided daily dosing schedule to a single daily dosing schedule.
Product Formulations
The lithium product formulation may influence the risk of lithium ADRs. Figure 3 presents computer-simulated curves for lithium carbonate 600 mg po q24h in both a standard-release and slow-release dosage form (Lithobid®) from single dose data originally described by Cooper et al (1978). Because of its rapid absorption, a single oral dose of regular-release lithium carbonate 600 mg resulted in a peak serum level usually within 90 minutes. However, the peak serum lithium concentration for the slow-release formulation occurred between four to six hours following ingestion. The peak serum lithium concentration at steady-state for the regular-release formulation was 0.76 ±_0.14 mEq/l whereas the peak concentration for the slow-release formulation was 0.42 ± 0.09 mEq/l. This represents a 45% decrease in peak level.
Figure 3. Serum concentration versus time cure for standard-release and slow-release lithium dosage formulations.
The slow-absorption lithium formulations were developed in an attempt to decrease the ADRs associated with peak and rapidly rising serum concentrations as well as to increase compliance. Products in which absorption is slowed to the greatest extent such as Quilonum Retard (Thornhill 1978), Lithionit Duretter (Borg et al 1974), and Lipett C (Perrson 1974), produced the lowest incidence of tremor and nausea but caused diarrhea when ingested on an empty stomach due to lithium reaching the large intestine and acting as an osmotic cathartic (Jeppsson and Sjogren 1975). Additionally, it has been noted that the urine concentrating ability of the kidneys is significantly better with slow-release lithium than the regular-release formulation) (Wallin and Alling 1979). On the other hand, another study was unable to illicit any differences in the adverse effect profiles of patients given both regular-release and slow-release (Lithobid) (Lyskowski and Nasrallah 1981). Thus, if slow-release lithium preparations do cause fewer ADRs, the difference in incidence is not large and is restricted to a minority of patients.
ADJUSTMENT OF LITHIUM DOSAGE
Lithium obeys first-order linear pharmacokinetics. The following formula describes the pharmacokinetic behavior of a first-order drug.
CSSav = (F) (DM) / (Cl) (lambda) (Eq. 1)
where C is the steady-state serum lithium concentration; F, the fraction of dose absorbed, lambda the dosage interval; Cl, renal clearance; and DM, the maintenance dose (Notari 1975).
In normal clinical situations, the only terms in the equation that change with an increase or decrease in dose are the maintenance dose and the lithium steady-state concentration. Thus, with all other terms being constant, the equation can be rearranged to read:
CSSav = (K) (DM) (Eq. 2)
Thus, as the dose increases or decreases, the steady-state serum lithium concentration must increase or decrease proportionally. Thus, a patient with a steady-state concentration of 1.0 mEq/l at 1200 mg/day will have a steady-state level of 1.25 mEq/l if the dose is increased to 1500 mg/day, whereas if the dose is decreased to 900 mg/day, the concentration in this equation is the mean serum level, not the peak, 12-hour, or trough level. However, clinically this method can be utilized as a reasonable prediction of the 12-hour steady-state lithium concentration.
Prospective Dosing Adjustments
Single Point Method - Dose Predictions. Cooper and Simpson (1976) described a method where an individual daily lithium dose could be predicted from a serum sample collected 24 hours after a single 600 mg test dose of lithium carbonate. The predicted dose was calculated to produce a steady-state serum lithium concentration between 0.6 and 1.2 mEq/l. Table 2 lists the dosages required to achieve a serum lithium level of 0.6-1.2 mEq/l using their method.
Cooper's method requires the patient be lithium-free prior to the test and the laboratory to be capable of accurate, reproducible lithium analyses. The authors indicated since the prediction technique was instituted, less frequent blood sampling were required and medication changes were rarely needed.
Naiman et al (1981) evaluated the practicality of Cooper's method. Of 13 patients who followed this dosing protocol, 4 (31%) failed to achieve the defined steady-state range. It was postulated that factors such as a lack of research laboratory facilities and drug interactions make the use of this dosing protocol impractical for the clinical psychiatrist. However, Gengo et al (1980) found the 0.6-1.2 mEq/l range was achieved in 23 of 24 patients dosed with this schedule.
Because of the two above conflicting reports Perry et al (1983) re-examined the relationship between single point lithium concentration and lithium maintenance dose. They replicated the findings of Cooper et al (1976). In addition, more confident relationships were noted when steady-state range concentrations for lithium prophylaxis (0.41-0.89 mEq/l) and for treatment of acute mania (0.9-1.27 mEq/l) were analyzed separately. Maintenance dose predictive equations were generated for these lithium steady-state ranges from a 24-hour single point level following a 1200-mg test dose. The dose schedules for these two relationships are given in Tables 3 and 4. The clinical usefulness of the new relationships was subjected to prospective testing (Perry et al 1984). A total of 20 patients were dosed utilizing the schedule for acute manics (Table 3) and 18 patients were dosed with lithium utilizing the prophylactic dosing schedule (Table 4) . Fifteen (83%) of the prophylactically dosed patients fell within the therapeutic range while eighteen (85%) of the acutely manic patients were within the therapeutic range.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TABLE 3: DOSAGES REQUIRED FOR 0.41-0.89 mEq/l SERUM LITHIUM LEVEL FOR RECURRENT AFFECTIVE DISORDER PROPHYLAXIS FOLLOWING A 1200 MG LITHIUM CARBONATE TEST DOSE (PERRY ET AL 1983). |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TABLE 4: DOSAGES REQUIRED FOR 0.9-1.3 mEq/l SERUM LITHIUM LEVEL FOR ACUTE MANICS FOLLOWING A 1200 MG LITHIUM CARBONATE TEST DOSE (PERRY ET AL 1983). |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Single Point Method - Steady-State Level Prediction.
A mathematical relationship between drug concentration in serum or plasma at steady state and a single drug concentration at some time after a test dose has been described by Slattery et al (1980). The derivation suggests that a direct proportional relationship exists between the mean steady state drug concentration and the initial dose drug concentration at some time. To investigate this mathematical model, the steady-state lithium serum concentrations observed at 12 hours post dose were retrospectively adjusted to an 1800 mg/d lithium maintenance dose for seventeen patients by Perry et al (1986). The normalized steady-state serum lithium concentrations were subjected to linear regression analysis utilizing the 24-hour serum lithium level observed after an initial 1200 mg lithium dose. The following straight line equation was derived where
CSS12h = 0.13 + 3.3 (C*24) (Eq. 3)
where C is the steady-state lithium concentration at 12h post-dose for a 1800 mg/day dose and C24h is the 24h serum lithium concentration following an initial 1200 mg dose. A correlation coefficient of 0.77 (p < 0.01) was calculated for this relationship. The equation was utilized in predicting steady-state serum lithium concentrations in 28 patients. The dosing nomogram in figure 4 can be utilized in place of the equation. Comparison of the observed and predicted serum lithium steady-state concentrations by linear regression analysis demonstrated a significant correlation (r2 = 0.64, p < 0.01). Slattery et al (1980) noted that the larger the range of half-lives for the population the poorer the correlation between the steady state and single dose concentrations. However, a four-fold variation in half-life introduces relatively little error. If one assumes a range of 10 to 40 hours for lithium half-lifes it is important that patients with significantly reduced renal function not be dosed with this equation because of their extended half-lifes which may exceed 40 hours.
Multiple Point Method - Steady-State Level Prediction.
A pharmacokinetic method that keeps the number of serum lithium samples to a minimum and more accurately predicts steady state lithium levels would be useful in facilitating the treatment of manic-depressive patients.
Perry et al (1982) have established a method whereby the steady-state serum lithium concentration can be predicted with somewhat greater precision than the single point method. The patient is administered a 1200 mg test dose of lithium carbonate. Serum lithium samples are obtained over a 24 hour period beginning at 12 hours following the dose. Usually only a 12- and 24-hour sample are necessary. These two data points allow the calculation of the elimination rate constant, ke. Since ke is available the accumulation factor R can be calculated according to equation 4.
R = 1 / 1 - eke lambda (Eq. 4)
where ke is the elimination rate constant and _ is the dosing interval produced by a 1200 mg/day dose. At this point the steady-state concentration for the 1200 mg test dose can be determined for the desired dosing interval according to equation 5.
CSS 12h TD = C*12h (1) (R) (Eq. 5)
Calculation of the steady-state lithium concentration for any lithium carbonate maintenance dose can be accomplished by solving equation 6.
DM / TD = CSS 12h MD / CSS 12h TD (Eq. 6)
The following five-step worksheet gives an example of this method for a patient.
1. Test Dose = 1200 mg on 7/14
2. Data Collection
|
Date |
|
|
|
|---|---|---|---|
|
7/15 |
0600 |
12 |
0.60 |
|
7/16 |
0600 |
36 |
0.30 |
4. Accumulation factor (R)
R = 1 / 1 - e-0.0288 (12) = 3.42
5. Predicted CSS12h TD = C12h(1) . R = 0.6 x 3.42 = 2.05 mEq/L12h
at a maintenance dose of 1200 mg q 12 hours.
Obviously this dose is too high. Half this dose would be more appropriate, i.e., 600 mg q 12 hours would produce at steady-state level of 1.02 mEq/l. A total of 22 patients were dosed according to the multiple point method. The correlation coefficient (r) of observed versus predicted serum lithium levels for these patients was 0.96 (p<0.001).
Even though this lithium steady-state prediction is more complex than the other dosing protocols it is superior with respect to its accuracy, sensitivity, and the availability of the lithium half-life to the clinician.
The maintenance dose protocols presented previously guarantee serum lithium concentrations only over rather widely spaced ranges. The multiple point serum lithium concentration prediction protocol offers greater flexibility in that the clinician can administer any dose to achieve any desired therapeutic serum lithium level with 77-90% confidence that the observed serum lithium level will be within ±0.1 mEq/l of the predicted serum lithium level. Although not as precise, the single-point serum lithium concentration prediction nomogram described above yields predictions that are within ± 0.15 mEq/l for 73% of the predictions.
Although the multiple point serum lithium concentration prediction protocol is the most accurate protocol for predicting steady-state serum lithium concentrations it is also the most difficult to use. To simplify its use, the authors developed a pharmacokinetic interactive computer program. It was written in BASIC and is run on either an IBM personal computer, an Apple 2E computer, or any IBM PC compatible computer. The user enters the patient's name, the serum levels, and times plus the number of hours after the dose that future blood level measurements will be drawn. The user receives a hard-copy output that states the half-life, the number of days necessary to reach steady-state and the correlation coefficient for the fit of the one-compartment pharmacokinetic model. This is followed by a table of steady-state blood levels that would be expected to result from various doses and dosing schedules. The table is especially helpful in giving the clinician a feel for the interaction between dose, schedule and half-life. It also can transform steady-state serum lithium concentrations not drawn at 12 hours after the dose into 12 hour levels. The data output gives the clinician useful information for treating that specific patient as well as understanding the kinetics of lithium in general. The clinician who does not have access to either a minicomputer or a programmable calculator can still utilize the multiple point method employing a hand-held calculator with a natural log function.
Clinicians can obtain the program software or a listing of the computer program by writing either of the chapters' authors.
Prospective Dosing Caveat
A potential source of error in the prospective dosing protocol is intraindividual variability in lithium clearance. The relationship between steady-state drug concentration and clearance for first-order linear kinetic models is demonstrated by Equation 7:
CSS = (F) (DM) / (Cl) (lambda) (Eq. 7)
where CSS is the average serum concentration at steady state, F, the fraction of the dose absorbed; DM, the maintenance dose, C1, the clearance, and /lambda the dosing interval. The value of F can be assumed to be 1.0, while the maintenance dose and dosing interval are constant during the dosing period. Thus, the relationship between a steady-state lithium concentration and lithium clearance is demonstrated by Equation 8:
CSS = [1 / (Cl)] (ke) (Eq. 8)
Where ke is the constant (F) (DM)/lambda (Wagner et al 1965). Lauritsen et al (1981) found night-time lithium clearances to be 22% less than daytime lithium clearances. It was suggested that the mechanism for diurnal variation in lithium clearance resulted from the glomerular filtration rate being significantly lower at night when the patients were in a supine position and being higher during the day because the patients were erect. Thus the primary source of error in lithium prospective dosing methods originates from dynamic glomerular filtration rate changes stemming from changes in the hyperactivity of the manic patient. Thus, if a patient is administered a lithium test dose in a hyperactive state, i.e., sleeping and supine only 2 hours per night, and then the 12-hour steady state lithium level is measured a week later when the patient is no longer hyperactive, i.e., sleeping 8 hours per night, a significantly higher than expected steady state lithium level is a strong possibility. The reverse situation should hold true, also. Thus, the hyperactive manic patient will have the lithium level rise significantly once activity levels and sleep normalize, while the euthymic patient will have the lithium concentration decrease significantly if mania recurs.
Because of the dynamic relationship of steady-state serum lithium concentrations between manic and euthymic state, the utility of dosing schedules and nomograms is questionable in manic patients. The only means by which the clinician can recognize this dynamic state is by being aware of the patient's lithium half-life. Manic patients with short-half lives will eventually develop longer half-lives once they become euthymic.
Non-pharmacokinetic approaches to dosing
Zetin et al (1986) utilizing stepwise multiple linear regression analysis derived a mathematical formula for determining a lithium dose based on a number of dependent variables. The regression model is presented in equation 9.
Dose = 746.83 (level) + 92.01 (status)-74.73 (total) -10.08 (age)+147.8(sex) + 5.95(weight) + 486.8 (Eq. 9).
where dose is expressed in milligrams per day, level is the desired lithium concentration in mEq/l, TCA is 1 for concomitant antidepressant use of 0 for no antidepressant use, age is in years, sex is 0 for females and 1 for males, status is 0 if an outpatient and 1 if an inpatient, and weight is in kg. The coefficient of variation (r2) for the equation was 0.45. In 68% of cases, the computer predicted and actual doses agreed to within ± 300 mg/d; in 93% of the cases the predicted and actual doses agreed to within ± 600 mg/d.
Lesar et al (1985) also utilized multiple linear regression analysis to derive a lithium dosing equation. Their multiple regression model not requiring a laboratory value, i.e., serum creatinine for a creatinine clearance determination, was defined by equation 10:
Dose = 9.56 - 1.19 (sex) + 0.064 (weight, kg) - 0.021 (age, years)- 2.73 (depression) + 2.26 (state) + 0.035(Clcr) (Eq. 10)
where the sex is 1 for males and 2 for females, depression is 1 for absence and 2 for presence and/or tricyclic antidepressant use, state is coded 1 for acute and 2 for non-acute symptoms, and Clcr is the creatinine clearance calculated by the Cockroft-Gault equation (Cockroft and Gault 1970) presented in equation 11:
Clcr = (140 - age)(weight, kg)/(72)(serum creatinine) (Eq. 11).
The resulting coefficient of variation (r2) was 0.45 (p<0.05). The Lesar equation predicted the actual doses required somewhat better than the Zetin equation. The authors recommended that the Zetin equation be limited to use in patients with creatinine clearances greater than 45 ml/min. It is assumed this is not a problem with the Lesar equation since creatinine clearance is accounted for in the equation.
Lithium Loading
Kook et al (1985) demonstrated the feasibility of using lithium loading doses to accelerate dosing. Thirty-eight patients were loaded with a dose of lithium carbonate (Lithobid) 30 mg/kg with a ceiling dose of 3000 mg. The loading dose was divided into three equal doses being administered at 4, 6, and 8 PM. For the males the levels at steady state ranged from 0.58 to 1.1 mEq/L while the females ranged from 0.45 to 1.4 mEq/L. However, only 39% of the females had levels that exceeded 0.9 mEq/L in contrast to 50% of the males. Three of the four women with high outlier levels were obese. Thus the authors recommended that ideal body weight (IBW) be utilized in the calculation of the loading dose for at least the women. For males IBW (kg) = 2.3 (inches> 60-) + 50 whereas for females IBW (kg) = 2.3 (inches> 60-) + 45. No patient experienced any ADRs during the loading procedure or in the 12 hours after loading when the lithium level was drawn. Keck et al (2001) utilized a loading and continuation dose of 20 mg/kg/d for 10 days in 15 acutely manic patients (7 males). The 12 hour mean serum lithium concentrations at day 2 were 0.7 mEq/L, at day 3 were 0.9 mEq/L, at day 4 were 1.0 mEq/L, and at day 7 were 1.1 mEq/L. Seven of the patients were well enough to be discharged before the 10-day observation period was completed. One patient was discontinued because of bradycardia. Otherwise the adverse effects were characterized as mild. Thus, at least when using this lithium loading strategy, the drug does not need to be titrated. Instead, a loading dose of 30 mg/lg on day 1 could be reasonably be followed by a continuation dose of 20 mg/kg/d given on a twice daily dosing schedule.
Conclusions
Pharmacokinetic and non-pharmacokinetic approaches to lithium dosing can be compared using the r2 values for the two approaches. The r2 value refers to the proportion of variation explained by a regression model. Thus, a model with r2 = 0.68 is said to explain 68% of the variability in the relationship between the dependent and independent variables. The coefficients of variation for the pharmacokinetic methods of Perry, i.e., multiple-point method and one-point method were 0.92 and 0.59, respectively, while the r2 values for the non-pharmacokinetic methods of Zetin and Lesar were 0.45 and 0.45, respectively. Although it is the most difficult method to use, the Perry multiple point method is the most accurate procedure to determine a desired lithium dose. It provides the clinician with the most information for dosing the patient in the future as opposed to the other methods. Additionally, Browne et al (1989) compared empirical lithium dosing to Perry's one-point method and Zetin's multivariate regression method. The Perry method produced a significantly greater percentage of clinically acceptable predictions compared with the population-based Zetin multivariate regression model.
VALPROATE
Valproate is supplied as either a free acid (valproic acid) in capsules or as a sodium salt (sodium valproate) in the syrup. Divalproex (Depakote) is an enteric coated formulation that is a 50:50 mixture of sodium valproate and valproic acid. The drug has a half-life ranging from 8 to 12 hours. The volume of distribution ranges from 0.1 to 0.4 L/kg with a mean of 0.25 L/kg. The fraction of the dose absorbed ranges from 0.9 to 0.95. Generally the generic forms of the drug are dosed tid or qid whereas the enteric coated formulation is dose on a bid to tid schedule. Valproate blood levels should be routinely drawn in the morning 12 hours after the last dose because of a circadian rhythm in clearance of the drug (Baselt 2000).
Dosing
Keck et al (1993) were the first to attempt to utilize a loading dose strategy for valproate in bipolar manic patients. Nineteen patients were loaded with valproate 20 mg/kg/d. On day 2, a 12-hour post dose blood sample was obtained and assayed. The dose was linearly adjusted such that the blood level exceeded 50 mcg/ml. Fifteen of the 19 patients who completed the protocol had a valproate level of 89 ± 19 mcg/ml by day 2. Adverse effects included mild sedation (3), moderate sedation (1), and nausea (1). The four dropouts were because of worsening of mania rather than adverse effects. Although 10 of the patients had ³ 50% decrease in the Young Mania Rating Scale (YMRS) only 3 of the patients were discharged on valproate alone.
Hirschfeld et al (1999) contrasted a more aggressive valproate loading dose procedure to empirical valproate dosing and lithium treatment. Bipolar manic patients (n=59) were randomized to one of three treatments. (1) divalproex 30 mg/kg/d for 2 days and then 20 mg/kg/d for 8 days; (2) divalproex 250 mg po tid for 2 days and then titrated to a valproate blood level of > 50 mcg/ml; or (3) lithium 300 mg po tid for 2 days and titrated to < 0.8 mEq/L for 8 days. By day 3, 84% (mean = 84 mcg/ml) of the loaded patients had valproate levels > 50 mcg/ml whereas only 30% (mean = 42 mcg/ml) of the empirically dosed patients achieved this level. Despite the rapidly achieved blood levels, at day 10 there was no difference in the final YMRS change scores. Additionally, there were no differences in the frequencies or types of adverse effects.
Bowden et al (1994) attempted to discern if a valproate loading dose strategy would accelerate the onset of response for valproate versus lithium. A total of 179 acutely manic inpatients were randomized to divalproex, lithium or placebo in a 2:1:2 ratio. Divalproex was dosed as 750 mg on day 1 and 1000 mg on day 2. A 12-hour post dose blood level was drawn and the dose was adjusted so as not to exceed 150 mcg/ml. At day 8 the mean valproate concentration was 77 mcg/ml and by day 21, 93 mcg/ml. Unfortunately the loading dose did not result in any difference in onset of response between divalproex and lithium.
Two studies have calculated the length of hospital stays in bipolar patients from uncontrolled reports. The number of hospital days for patients treated with valproate 20 mg/kg/d as a starting dose was compared to lithium. Before these data are taken too seriously it must be remembered that overall for the two reports only 38% of the lithium treated patients had levels above 1.0 mEq/L. Thus it would be safe to assume that many of the patients had subtherapeutic lithium levels which would certainly bias the outcome of these findings in favor of valproate. Frye et al (1996) reported a mean stay of 10 days for divalproex versus 18 days for lithium. The two treatment groups were similar in the amount of adjunctive treatment, illness severity, and number of hospitalizations. Keck et al (1996) likewise reported a mean stay of 14 days for divalproex versus 18 days for lithium. The annual direct health costs for the valproate treated patients were 9% less than for the lithium patients. However, the entire cost savings was a function of the initial hospitalization, i.e., the four extra days of hospitalization.
Adverse Effects
The best contrast of adverse effects between valproate and lithium is available from a 12 month prophylactic efficacy study in 375 bipolar I patients conducted by Bowden et al (2000). Divalproex ADRs that occurred more often than either lithium or placebo in > 5% of patients included sedation (42%), weight gain (21%), alopecia (16%), infection (27%), and tinnitus (6%). Lithium ADRs that occurred more often than either divalproex or placebo included diarrhea (46%), polyuria (8%), and thirst (6%). Tremor occurred more often than placebo with both divalproex (41%) and lithium (42%).
Bowden et al (1996) treated 65 manic patients with divalproex, 750 mg/day for 2 days and then 1,000 mg/day for 2 days. The dose was then increased as clinically indicated, but not to exceed 150 mcg/ml for the remainder of the 21-day study, unless patients developed adverse effects. Those patients with valproate levels >125 mcg/ml experienced a greater incidence of nausea (27.3%), vomiting (18.2%) and sedation (18.2%). Thus if patients develop significant gastrointestinal ADRs on the enteric coated formulation the blood level ought to be ascertained to determine whether lowering the dose is indicated.
The non-enteric coated formulation of valproate is associated with gastrointestinal ADRs that anorexia, indigestion, heartburn, nausea, vomiting, and transient diarrhea in a minority of patients. However, the less expensive generic formulation is well tolerated by 75% of patients. To avoid or minimize the gastrointestinal ADRs it is recommended that administer the drug food, titrate the dose of the drug up at a slower rte, add an histamine-2 blocker or utilize the enteric-coated or sprinkle formulations that are 6-times more expensive. Switching to the enteric-coated formulation (Depakote) is successful in avoiding gastrointestinal ADRs in 85% of patients (Zarate et al 1999). However, only 29% of patients have gastrointestinal ADRs from the non-enteric coated formulation.
Therapeutic Drug Monitoring
Bowden et al (1996) treated 65 manic patients with divalproex, 750 mg/day for 2 days and then 1,000 mg/day for 2 days. The dose was then increased as clinically indicated, but not to exceed 150 mcg/ml for the remainder of the 21-day study, unless patients developed adverse effects. Table 5 demonstrates the distribution of bipolar patients experiencing a ³ 20% decrease in the manic symptom scale at day 5 of treatment. Valproate levels ³ 45 mcg/ml were associated with a 68% response rate versus a 20% response rate for doses producing lower blood levels. An important caveat regarding this study is that it was not a fixed dose study. Thus the findings need to be replicated in a fixed dose treatment study to validate these findings.
|
Response VPR (mcg/ml) |
Responders |
Nonresponders |
|
> 45 |
34 |
16 |
|
< 45 |
3 |
12 |
Perry et al (2000) examined the relationship between valproate plasma concentrations and the independent variables of gender, dose, smoking status, serum albumin, AST, and LDH in 48 bipolar patients being maintained on divalproex. The best correlation between these variables was plasma valproate (mcg/ml) = 1.8 (mg/kg/d-dose) + 14.7 (serum albumin) - 15.7 (F = 9.634, p = 0.0003, r2 = 0.30). Thus a bipolar patient taking divalproex 20 mg/kg/d with a serum albumin of 4 g% is predicted to have a steady valproate level of 79 mcg/ml.
CARBAMAZEPINE
A single dose of carbamazepine has a half-life that ranges from 25-65 hours. However, because of auto-induction of the hepatic enzyme CYP4503A4, the multiple dose half-life ranges from 12-17 hours. Induction begins within a week and maximizes at about 3 weeks. De-induction of CYP4503A4 requires a similar period of time. Carbamazepine blood level samples are usually drawn in the morning 12 hours after the last dose. When switching patients from a brand name ot a generic product for from one generic to another it is recommended to monitor the carbamazepine levels weekly for 2 weeks.
Simhandl et al (1993) recommend that a carbamazepine-high serum level (6.6-9.4 ug/ml) or lithium level between 0.6-0.8 mEq/L is preferred to a low carbamazepine serum level (3.5-5.9 ug/ml) in the prophylaxis of UAD. There were no differences between groups in the efficacy of the bipolar patients.
REFERENCES
Abou-Saleh MT, Coppen A (1989). The efficacy of low-dose lithium: clinical, psychological, and biological correlates. J Psychiat Res 23:157-62.
Amdisen A (1977). Serum level monitoring and clinical pharmacokinetics of lithium. Clin Pharmacokinet 2:73-92.
Baselt RC (2000). Disposition of Toxic Drugs and Chemicals in Man, 5th edition. Chemical Toxicology Institute, Foster City, CA.
Borg KO, Jeppsson J, Sjogren J (1974). Influence of the dissolution rate of lithium tablets on side effect. Acta Pharm Sued 11:133-140.
Bowden CL, Brugger AM, Swann AC, et al (1994). Efficacy of divalproex vs lithium and placebo in the treatment of mania. JAMA 271:918-24.
Bowden CL, Janicak PG, Orsulak P, et al (1996). Relation of serum valproate concentration to response in mania. Am J Psychiatry 153:765-770.
Bowden CL, Calabrese JR, McElroy SL, et al (2000). A randomized, placebo-controlled 12-month trial of divalproex and lithium in treatment of outpatients with bipolar I disorder. Arch Gen Psychiatry 57:481-9.
Browne JL, Huffman CS, Golden RN (1989). A comparison of pharmacokinetic versus empirical lithium dosing techniques. Ther Drug Monit 11:149-54.
Cockroft DW, Gault MH (1970). Prediction of creatinine clearance from serum creatinine. Nephron 16:31-41.
Cooper TB, Simpson GM (1976). The 24-hour serum lithium level as a prognosticator of dosage requirements: a 2-year follow-up study. Am J Psychia-try 133:440-43.
Cooper TB, Simpson GM, Lee H, Bergner PE (1978). Evaluation of a slow-release lithium carbonate formulation. Am J Psychiatry 135:917-22.
Coppen A, Abou-Saleh M, Milln P, Wood K (1983). Decreasing lithium dosage reduces morbidity and side effects during prophylaxis. J Affective Disord 5:353-62.
Frye MA, Altshuler LL, Szuba MP, et al (1996). The relationship between antimanic agent for treatment of classic or dysphoric mania and length of hospital stay J Clin Psychiatry. 57:17-21
Gelenberg AJ, Kane JM, Keller MB, et al (1989). Comparison of standard and low serum levels of lithium for maintenance treatment of bipolar disorder. N Eng J Med 321:1489-93.
Gengo F, Timko J, D'Antonio J, et al (1980). Prediction of dosage of lithium carbonate use of a standard predictive method. J Clin Psychiatry 41:319-21.
Hirschfeld RMA, Allen MH, McEvoy JP, et al (1999). Safety and tolerability of oral loading divalproex sodium in acutely manic bipolar patients. J Clin Psychiatry 60:815-8.
Hullin RP (1979). Minimum effective plasma lithium levels for long-term preventive treatment of recurrent affective disorders. In: Copper TB, Gershon S, Kline NS, et al, eds. Lithium: Controversies and unresolved issues. Amsterdam: Excerpta Medica 333-34.
Jeppsson J, Sjogren J (1975). The influence of food on side effects and absorption of lithium. Acta Psychiatr Scand 51:285-8.
Keck Jr.PE, McElroy SL, Tugrul KC, et al (1993). Valproate oral loading in the treatment of acute mania. J Clin Psychiatry 54:305-308.
Keck Jr.PE, Nabulsi AA, Taylor JL, et al (1996). A pharmacoeconomic model of divalproex vs lithium in the acute and prophylactic treatment of bipolar I disorder. J Clin Psychiatry 57:213-22.
Keck Jr PE, Strakowski SM, Hawkins JM, et al (2001). A pilot study of rapid lithium administration in the treatment of mania. Bipolar Disorders 3:68-72.
Keller MB, Lavori PW, Kane JM, et al (1992). Subsyndromal symptoms in bipolar disorder: a comparison of standard and low serum levels of lithium. Arch Gen Psychiatry 49:371-6.
Kook KA, Stimmel GL, Wilkins JN (1985). Accuracy and safety of a priori lithium loading. J Clin Psychiatry 46:49-51.
Lauritsen BJ, Mellerup ET, Plenge P, et al (1981). Serum lithium concentrations around the clock with different treatment regimens and the diurnal variation of the renal lithium clearance. Acta Psychiatr Scand 64:314-9.
Lesar TS, Tollefson GK, Koch M (1985). Relationship between patient variables and lithium dosage requirements. J Clin Psychiatry 46:133-36.
Lyskowski J, Nasrallah HA (1981). Slowed-release lithium: a review and a comparative study. J Clin Psychopharmacol 1:406-08.
Maj M (1990). Clinical prediction of response to lithium prophylaxis in bipolar patients: the importance of the previous pattern of course of the illness. Clin Neuropharmacol 13(suppl 1):S66-70.
Mellerup ET, Dam H, Plenge P, Widding A, Rafaelsen OJ (1985). Serum lithium minimum and diuresis. Psychiatry Res 14:309-13.
Naiman IF, Muniz CE, Stewart RB, Yost RL (1981). Practicality of a lithium dosing guide. Am J Psychiatry 138:1369-71.
Notari RE (1975). Biopharmaceutics and Pharmacokinetics. New York: Marcel Dekker pp. 167-168.
Perry PJ, Alexander B, Dunner FJ, Schoenwald RD, Pfohl BM, Miller D (1982). A single dose kinetic protocol for predicting steady state lithium levels. J Clin Psychopharmacol 1:114-18.
Perry PJ, Alexander B, Prince RA, Dunner FJ (1986). A single point dosing protocol for predicting steady state lithium levels. Br J Psychiatry 148:401-05.
Perry PJ, Alexander B, Prince RA, et al (1984). Evaluation of two prospective lithium maintenance dose schedules. J Clin Psychopharmacol 4:242-46.
Perry PJ, Prince RA, Alexander B, et al (1983). Prediction of lithium maintenance doses using a single point prediction protocol. J Clin Psychoph-armacol 3:13-17.
Perry PJ, Bever-Stille KA, Arndt S, et al (2000). Valproate plasma concentrations to dose correlations in bipolar affective disorder. J Clin Psychopharmaco 20:277-9.
Persson G (1974). Plasma lithium levels and side effects during administration of a slow-release lithium sulfate preparation (lithium Lipett C) and lithium carbonate tablets. Acta Psychiatr Scand 50:174-182.
Prien RF, Caffey EM, Klett CJ (1971). Relationship between serum lithium level and clinical response in acute mania treated with lithium. Br J Psychiatry 120:409-14.
Simhandl C, Denk E, Thau K (1993). The comparative efficacy of carbamazepine low and high serum level and lithium carbonate in the prophylaxis of affective disorders. J of Affective Disorders 28:221-231.
Slattery JT, Gibaldi M, Koup JR (1980). Prediction of maintenance dose required to attain a desired drug concentration at steady state from a single determination of concentration after an initial dose. Clin Phar-macokinet 5:377-85.
Solomon DA, Ristow R, Keller MB, et al (1996). Serum lithium levels and psychosocial function in patients with bipolar I disorder. Am J Psychiatry 153:1301-1307.
Stokes PE, Kocsis JH, Arcuni OJ (1976). Relationship of lithium chloride dose to treatment response in mania. Arch Gen Psychiatry 33:1080-4.
Thornhill DP (1978). Comparison of ordinary and sustained-release lithium carbonate in manic patients. Br J Clin Pharmacol 5:352P.
Thornhill DP, Field SP (1982). Distribution of lithium elimination rates in a selected population of psychiatric patients. Eur J Clin Pharmacol 21:351-4.
Vestergaard P, Wentzer LR, Brodersen A, et al (1998). Outcome of lithium prophylaxis: A prospective follow-up of affective disorder patients assigned to high and low serum lithium levels. Acta Psychiatrica Scand 98: 310-5.
Wagner JG, Northam JL, Alway CD, et al (1965). Blood levels of drugs at the equilibrium state after multiple dosing. Nature 207:1301-02.
Wallin L, Alling C (1979). Effect of sustained-release lithium tablets on renal function. Br Med J 2:1332.
Zarate CA, Tohen M, Narendran R, et al (1999). The adverse effect profile and efficacy of divalproex sodium compared with valproic acid: A pharmacoepidemiology study. J Clin Psychiatry 60:232-6.
Zetin M, Garber D, DeAntonio M, et al (1986). Prediction of lithium dose: a mathematical alternative to the test dose method. J Clin Psychiatry 97:175-8.
