©1993-2007

Diabetes Insipidus: Drug Induced

Jim P. Getzinger, M.D.

Diabetes insipidus is the excretion of a large volume of hypotonic, insipid (tasteless) urine, usually accompanied by excessive polydipsia. There are three pathophysiologic mechanisms in the differential diagnosis of diabetes insipidus:

Hypothalamic diabetes insipidus (or neurogenic DI) is the inability to secrete (and usually to synthesize) vasopressin in response to increased osmolality. There is no concentration of the dilute filtrate in the renal collecting duct, and a large volume of urine is excreted. This produces an increase in serum osmolality with stimulation of thirst and secondary polydipsia. Levels of vasopressin in plasma are unmeasurable or low.   Common mechanisms for this are stroke, infection, tumor, trauma, and systemic diseases.

Nephrogenic diabetes insipidus is a disorder in which the otherwise normal kidney is unable to respond to vasopressin. As in hypothalamic diabetes insipidus, the dilute filtrate entering the collecting duct is excreted as a large volume of hypotonic urine. There is a rise in serum osmolality that stimulates thirst and produces polydipsia. Unlike hypothalamic diabetes insipidus, however, measured levels of vasopressin in plasma will be high.  Causes of NDI include congenital disease, renal failure, sickle cell anemia, hypercalcemia, hypokalemia, and certain drugs.

Primary polydipsia is a primary disorder of thirst stimulation. Ingested water produces a mild decrease in serum osmolality that turns off secretion of vasopressin. In the absence of vasopressin action on the kidney, there is lack of concentration of urine and excretion of a large volume. Measured vasopressin in plasma is low. While the pathophysiologic mechanisms for the three disorders are distinct, patients in each category usually have polyuria, polydipsia, and normal serum sodium. This is because the normal thirst mechanism is sufficiently sensitive to maintain fluid balance in the first two disorders, and the kidney is normally sufficiently responsive to excrete the water load in the third.

Renal insensitivity to the antidiuretic effect of vasopressin results in the excretion of an increased volume of dilute urine, a decrease in body water, and a rise in plasma osmolality, which, by stimulating the thirst mechanism, induces a compensatory increase in water intake. As a consequence, the osmolality of body fluid stabilizes at a new higher level that approximates the osmotic threshold for thirst. The magnitude of polyuria and polydipsia varies greatly depending on a number of factors, including individual differences in solute load, in the set-point and sensitivity of the thirst and vasopressin osmostats, and in the degree of renal insensitivity to vasopressin. It is important to note that the renal insensitivity to vasopressin need not be complete for polyuria to occur.  It is necessary only that the defect be great enough to prevent concentration of the urine at plasma vasopressin levels achievable under ordinary conditions of ad libitum water intake (i.e., at plasma osmolalities near the osmotic threshold for thirst). Calculations analogous to those used for states of vasopressin deficiency indicate that this requirement is not met until the renal sensitivity to vasopressin is reduced by more than 10-fold (i.e., urine concentration does not occur until plasma vasopressin levels exceed 5 pg/mL). Studies of the relationship between urine osmolality and plasma vasopressin in patients with familial and acquired forms of partial nephrogenic diabetes insipidus are consistent with these calculations.  Because renal insensitivity to the hormone is so often incomplete, many patients with nephrogenic diabetes insipidus are able to concentrate their urine when they are deprived of water  or given large doses of vasopressin.

Drug-induced NDI (DNDI) is a serious side effect of various drugs that has been most frequently observed in patients treated with lithium, demeclocycline, and methoxyflurane. In addition, other drugs, including amphotericin B, bentamicin, fluoride, phenytoin, aminoglycosides, propoxyphene, and several cancer treatment drugs have been reported to similarly impair the kidneys' urine-concentrating ability.

Nephrogenic diabetes insipidus (NDI) can be induced in people by giving them lithium over a period of time. In fact, 5% to 20% of patients receiving lithium as part of their therapy develop NDI. NDI symptoms may disappear in as little as three weeks after lithium use is stopped, though it may take up to a year for the NDI symptoms to resolve. Lithium is believed to inhibit the production of an important metabolic regulator called cAMP in the kidneys' collecting tubule cells. These cells normally help balance the kidneys' ability to reabsorb water. The antidiuretic hormone, vasopressin (VP) is instrumental in cAMP production. It also increases prostaglandin E₂ (PGE₂), which acts as a hormone that helps maintain a balanced kidney function by stimulating excretion of sodium in the urine and increasing the excretion of urine in general.

Lithium upsets balanced kidney function by increasing the PGE₂ levels and decreasing cAMP levels. Indomethacin seems to curtail the production of PGE₂, and this is followed by increased kidney cAMP production. Tests on people without NDI show indomethacin promotes sodium and water  reabsorption by preventing prostaglandin-mediated sodium loss.

Lithium also interferes with fluid and sodium balance by inhibiting the expression of the water-transporting protein, aquaporin2 (AQP2). AQP2 takes a long time to return to normal levels of expression after lithium use is stopped, and this may lead to prolonged symptoms of LINDI.  Aquaporin2,  when signaled by the binding of ADH and the vasopressin-2 receptor, inserts itself into the principal cells of the kidney collecting duct in order to make them much more water permeable than normal. This enhanced water permeability enables the principal cells to reabsorb water and concentrate urine.

Delayed nephrotoxicity secondary to fluoride resulting from metabolism of methoxyflurane and possibly enflurane may occur.  Alternatively, fluoride may produce intrarenal vasodilatation with increased medullary blood flow, which interferes with the countercurrent mechanism necessary for optimal concentration of the urine.

Demeclocycline is a drug normally used to treat Syndrome of Inappropriate ADH (SIADH).  It acts to inhibit the action of ADH on the distal tubule for reabsorption.  If a patient is erroneously diagnosed with SIADH and given this medication, they could develop NDI.

The drug ifosfamide is used in the treatment of various tumors in adults and children. However, it can have significant side-effects such as bleeding and inflammation of the urinary bladder and kidney poisoning that may include tubular and glomeruli impairment. Negro, et al., report on a 48-year-old woman with breast cancer who was being treated with ifosfamide. She developed ifosfamide-induced Fanconi syndrome along with associated NDI. NDI has been reported in other patients treated with ifosfamide, though the authors note that in all cases of ifosfamide-induced Fanconi syndrome, diminished urine concentrating ability (such as is associated with NDI) may be partially caused by low blood levels of potassium.

In most cases, once the drug is discontinued (unless lithium has been used long-term) or the electrolyte disorder is balanced, the nephrogenic DI can markedly improve and often resolve completely.  Management of the sequelae of this disorder though, depends on the basis of its cause.

The Hypernatremic Patient

The goal of treating hypernatremia in patients with diabetes insipidus is the restoration of normal plasma volume and tonicity. Desmopressin should be injected to maintain a concentrated urine. If there are circulatory disturbances due to hypovolemia, isotonic saline should be given until systemic hemodynamics is stabilized. In fact, isotonic saline is relatively hypotonic to plasma in patients with severe hypernatremia and simultaneously corrects both volume and water deficits. After volume deficits are corrected, the hypernatremia should be treated intravenously with 5% dextrose in water. Water can be given by mouth if the patient is conscious and able to drink.

The water deficit in these patients can be calculated on the basis of the serum sodium concentration and on the assumption that 60% of body weight is water. For example, if the patient's usual weight is 75 kg, total body water would normally be 75 kg × 0.6 = 45 liters. If the serum sodium value is 154 mEq per liter, the patient has a 10% deficit of water--(154 - 140) ÷ 140--and requires 4.5 liters of water to correct the deficit. Obviously, continuing losses of water must also be replaced. Despite inaccuracies, including the assumption that body water is always 60% of the body weight and the postulate that water is lost uniformly throughout all body cells, this approach provides an approximate value that can be used in planning therapy. The major problem is to determine the appropriate rate at which to lower serum sodium concentration to normal, although there are few hard data on how rapidly serum sodium concentration can be lowered safely in patients with hypernatremia. Because seizures or even fatal cerebral edema may occur when serum sodium concentration is lowered rapidly, the best estimate is that hypernatremia should be corrected over 48 to 72 hours and at a rate not to exceed 0.5 to 2.0 mEq per liter per hour. As total body water expands, the serum sodium concentration may fall proportionately. Serum electrolyte values should be monitored frequently to ensure an appropriate response.

Treatment of the hypernatremia due to water loss such as occurs in untreated patients with diabetes insipidus must also address associated electrolyte abnormalities and underlying medical and surgical conditions. A special example of this would be the patient with diabetes insipidus who has coexisting hyperglycemia. When marked hyperglycemia exists, the "corrected" serum sodium value should be used to calculate the water deficit. Slightly low or normal serum sodium concentrations in the presence of high serum glucose concentrations often result, when corrected, in hypernatremic values. The corrected serum sodium concentrations can be calculated by increasing the serum sodium concentration by 1.5 mEq per liter for every 100 mg per dL increment in the serum glucose concentration above 100 mg per dL. For example, in a patient with a sodium level of 138 mEq per liter and a glucose level of 700 mg per dL the corrected serum sodium would be 138 + (1.5 × 6), or 147 mEq per liter.

The Hyponatremic Patient

Hyponatremia in patients with diabetes insipidus occurs almost exclusively in patients who are overhydrated orally or parenterally while they are being treated with desmopressin. The severity of hyponatremia closely correlates with the extent of water overload. The magnitude of the excessive body water can be calculated by use of the same approach as described for hypernatremia. Occasionally, hyponatremia is aggravated by large amounts of sodium in the urine, probably related to increased levels of atrial natriuretic peptide, inhibition of aldosterone, and an increased glomerular filtration rate. Random measurements of urine sodium concentrations should be obtained to determine whether they are in the usual range (20 to 40 mEq per liter), high (about 100 mEq per liter), or extremely high (about 180 mEq per liter). The hyponatremia due to natriuresis in the water-overloaded patient can be corrected only partially with saline infusions, because the natriuresis continues until the hypervolemic state is corrected. Hyponatremia can be caused or aggravated by thyroid or adrenal insufficiency.

The symptoms of hyponatremia depend on the acuteness and degree of the condition. Neurologic manifestations of acute water intoxication are usually not observed until the plasma sodium concentration has fallen to less than 125 mEq per liter. They include nausea, emesis, muscular twitching, grand mal seizures, and coma. Acute water intoxication, which causes plasma sodium concentration to fall below 125 mEq per liter in less than 24 hours, carries substantial morbidity and mortality rates. When plasma sodium is lowered slowly to the same level, patients are usually less symptomatic.

There is a large body of literature on the appropriate rate at which to correct hyponatremia. Rapidly occurring (acute) and marked hyponatremia can be lethal and should be treated urgently. Under these conditions, and when neurologic symptoms are severe, initial therapy should raise the serum sodium concentration by 1 to 2 mEq per liter per hour regardless of the duration of the electrolyte abnormality. Most authorities agree that the rate of change in serum sodium concentrations should not exceed 12 to 20 mEq per liter per day. However, in patients with chronic hyponatremia, correction of serum sodium approximating this rate can occasionally cause serious, even fatal, complications by inducing the central pontine myelinolysis syndrome.

In the treatment of the asymptomatic mildly hyponatremic patient, fluid restriction is adequate. Urine should be analyzed every 4 to 8 hours for volume and osmolality, and fluid replacement should be ordered in relation to urine volume. Remember also that insensible fluid losses of about 600 mL of free water per day occur in the usual adult. It is inappropriate to write for a fixed amount of fluid replacement. Plasma sodium concentration should be checked frequently, and fluid replacement adjusted accordingly. The complaint of thirst by a water-restricted patient should never be ignored. Long-term management is usually less disrupted by adjusting fluid intake than by discontinuing hormonal therapy and allowing the patient to "break through." In severe situations, however, interruption of hormonal therapy may be justified with close monitoring. Alternatively, when the patient has symptomatic or severe hyponatremia (serum sodium value less than 115 mEq per liter in chronic hyponatremia or 125 mEq per liter in acute hyponatremia), the patient should be treated with intravenous furosemide and hypertonic saline (even in the presence of vasopressin, furosemide causes the excretion of urine that is slightly hypotonic or isotonic). After the injection of 40 mg or more of furosemide, 100 mL of 3% saline should be infused in the first hour. This rate should be decreased or discontinued subsequently if symptoms have ameliorated or if the plasma sodium concentration has increased by more than 2 mEq per liter in that hour. It is rarely necessary to infuse more than a total of 250 to 300 mL of 3% saline.

Some patients with nephrogenic diabetes insipidus can be treated by eliminating the drug or disease responsible for the disorder., however, the only practical form of treatment for many years has been to restrict sodium intake or to administer thiazide diuretics.  The latter generally reduce polyuria by about 50% and improve dehydration. Thiazides work by a vasopressin-independent mechanism that involves inhibition of Na⁺ reabsorption in the diluting segment of the nephron followed by increased reabsorption of glomerular filtrate in the proximal tubule. Treatment with prostaglandin inhibitors or  amiloride has been shown to be efficacious in some cases.

1. Behrman:  Nelson Textbook of Pediatrics, 15^th ed., © 1996 W.B. Saunders Company, pg. 1513-1514.

2. Brenner & Rector’s The Kidney, 5^th Ed.,  © 1996 W.B. Saunders Company, pg. 889-898.

3. Cecil’s  Textbook of Medicine, 19^th Ed., © 1992 W.B. Saunders Company, pg. 1223-1226.

4. Rakel:  Conn’s Current Therapy 1998, 50^th Ed., © 1998 W.B. Saunders Company, pg. 620-623.

5. Rosen:  Emergency Medicine:  Concepts and Clinical Practice, 4^th Ed., © 1998 Mosby-Year Book, Inc, pg. 2432-2434.

6. “Ifosfamide-induced Renal Fanconi Syndrome with Associated Nephrogenic Diabetes Insipidus in an Adult Patient,” Nephrology, dialysis, Transplantation, June, 1998.

7. Stoelting, R.K., Anesthesia and Co-existing Disease, 3rd Ed., pp. 294-297.