Biochemical Basis for the Hypercoagulable State Seen in Cushing Syndrome

Hypothesis  Cushing syndrome (CS) is associated with a hypercoagulable state that results in a 4-fold increase in the incidence of pulmonary embolism, deep venous thrombosis, and a 4-fold mortality rate compared with the general population. The incidence of CS in humans is approximately 2 to 5 per million per year, whereas in dogs it is much higher. The clinical complications of CS in humans are also manifested in dogs. We used a dog model of CS to better define the biochemical basis for the hypercoagulable state seen in the disease.

Design  A consecutive sample of dogs with CS and a cohort of healthy control dogs identified at a "well-dog check" were enrolled. All dogs underwent blood assays to identify the levels of procoagulant factors, natural antithrombotics, and the degree of ongoing activation of the coagulation cascade.

Setting  University veterinary medical teaching hospital.

Results  A total of 86 dogs were enrolled, 56 with CS and 30 control dogs. Levels of procoagulation factors II, V, VII, IX, X, XII, and fibrinogen were significantly increased in dogs with CS (P<.05). The natural antithrombotic antithrombin was significantly decreased in dogs with CS (P<.02). Thrombin-antithrombin complexes, a marker of subclinical thrombosis, were significantly increased in dogs with CS (P<.05).

Conclusions  The hypercoagulable state of CS is demonstrated by an increase in thrombin-antithrombin complexes. This hypercoagulable state may be caused in part by (1) an elevation of procoagulant factors, and (2) a decrease in antithrombin. Because of the similar clinical and biochemical changes between dogs with CS and humans, this canine model may be a useful tool for the future study of the hypercoagulable state in CS.

PATIENTS WITH Cushing syndrome (CS) have an accelerated mortality rate when compared with the general population. Thromboembolic complications are 4 times more frequent in CS than in the general population and contribute to this increase in mortality.¹

Several human studies have shown an increase in factor VIII in CS,^2-5 which seems to correlate with the level of hypercortisolism (CS) and returns to normal after treatment.^2-4 Other coagulation factors found to be increased include II, V, VII, IX, X, XI, and XII.^2-5 It has been postulated that because of these increases in procoagulant factors, patients with CS have a tendency toward thrombogenesis.

The incidence of CS in humans is approximately 2 to 5 per million per year, making it difficult to perform studies of large populations of humans. The incidence of naturally occurring CS in dogs, however, is much greater. Many of the biological changes seen in people with CS are also manifested in dogs, including thromboembolic complications.⁶ We found a significant increase in factors V, X, and plasminogen in a previous study⁷; however, the sample size of dogs was small and fibrinolytic factors were not measured. Therefore, this study was designed to further examine the molecular changes in coagulation that occur with CS.

All of the dogs in this study were patients at the Veterinary Medical Teaching Hospital, School of Veterinary Medicine, University of California–Davis, Sacramento, being seen for health maintenance or treatment. Dogs with CS had abnormalities typical of canine CS on history, physical examination, serum chemistry tests, urine cortisol-creatinine ratio, and adrenocorticotropic hormone stimulation test. The dogs were not followed up for thromboembolism as part of this study. Control dogs were aged 6 years or older and there were no complaints by owners, physical examination findings, or clinical pathology data to suggest underlying disease. Consent was obtained from each dog's owner. The study was approved by the investigation review board.

Dogs were fasted for 12 hours before blood was drawn. All blood samples were obtained before any other sample collections, procedures, or provocative tests. Blood was drawn by atraumatic jugular venipuncture into buffered sodium citrate (3.2%) tubes. Blood samples measured 5 mL. The samples were gently mixed, placed in vacuum test tubes, and transported in an ice bath. Samples were centrifuged at 3400 rpm for 10 minutes and 5 aliquots of 0.5-mL citrated plasma were placed into 5 plastic test tubes and stored at –70°C. Before analysis, samples were quick-thawed in a 37°C waterbath for 5 minutes and vortexed before use.

Assays were conducted on quick-thawed plasma. Samples were tested for coagulation factors II, V, VII, VIII, IX, X, XI, XII, von Willebrand factor, thrombin-antithrombin complex (TAT), antithrombin, fibrinogen, plasminogen, and plasminogen activator inhibitor 1. Factor assays were performed using Dade factor deficient substrate (Dade International Inc, Miami, Fla) on the MLA 1000C coagulation analyzer (Medical Laboratory Automation, Pleasantville, NY). A single calibration curve was run before sample testing. The calibrator (plasma sample) consisted of a pool of at least 10 normal dog samples. The pooled plasma defined normal canine activity for each factor. The von Willebrand factor was tested using an enzyme-linked immunoassay technique. Results are reported in international units per milliliter. In-house studies confirmed the cross-reactivity of this method to canine samples. Thrombin-antithrombin complex was tested using an enzyme-linked immunoassay method. The result is reported as µg/L. Previous references and in-house data confirmed the cross-reactivity of this method to canine samples. Antithrombin was measured using a factor Xa inhibition method on the MLA 1000C. This technique measures antithrombin activity. Results are reported by a functional assay as percentage of normal dog antithrombin activity. Fibrinogen levels were measured using the modified Clauss technique using Dade reagents on the MLA 1000C coagulation analyzer. The calibration technique was run before beginning the canine sample using Fibrinogen Reference Material (College of American Pathologists, Miami, Fla). Plasminogen was measured with a chromogenic assay using Dade reagents on the MLA 1000C coagulation analyzer. Results are reported as percentage of activity. Tissue plasminogen activator and plasminogen activator measure the quantitative activity in plasma. Tissue plasminogen activator is reported as international units per milliliter and plasminogen activator inhibitor 1 is reported asinternational units per milliliter, where 1 unit is defined as the amount of plasminogen activator inhibitor 1 that inhibits 1 international unit of 1-chain tissue plasminogen activator.

An analysis of variance was used to analyze the data. Mean values were compared between dogs with CS and control dogs. A P value less than .05 was used to determine a significant difference between means. The SE of the mean was used to evaluate variability within the groups.

A total of 86 dogs were enrolled, 56 with CS and 30 control dogs. Levels of procoagulation factors II, V, VII, IX, X, XII, and the von Willebrand factor were significantly increased in the dogs with CS (P<.05) (Figure 1). Differences seen in factors VIII and XI between dogs with CS and controls did not reach significance.

Thrombin-antithrombin complex, a marker of subclinical thrombosis, was noted to be significantly increased in dogs with CS compared with control dogs (P<.05) (Figure 2). Antithrombin was significantly decreased in dogs with CS compared with controls (P<.02). Differences in plasminogen, plasminogen activator inhibitor 1, and fibrinogen factor between dogs with CS and controls did not reach statistical significance (Figure 3).

Thromboembolic complications occur 4 times more frequently in patients with CS than in those without CS, and contribute to increased mortality. The incidence of deep venous thrombosis in nonhospitalized human patients is less than 1%, while the incidence of deep venous thrombosis in CS ranges from 7% to 25%.^1,2,8 The incidence of pulmonary embolism in the general population of hospitalized patients is approximately 1%; however, in patients with CS it has been estimated to be from 2% to 4% and is frequently the cause of death.^1,2,8 Dogs with CS suffer the same clinical manifestations as humans, including an increased incidence of thromboembolic events that frequently result in death.⁹

Chronic increases in serum cortisol levels cause the clinical manifestations of CS. Cortisol promotes gluconeogenesis, proteolysis, and lipolysis. It also has potent anti-inflammatory effects through the inhibition of prostaglandin synthesis, stabilization of lysosomal membranes, inhibition of interleukin 1 and interleukin 2 secretion, and a reduction in immunoglobulin synthesis. Based on the increased incidence of thromboembolic complications, cortisol may also have an effect on the hemostatic mechanism.³

Thrombin-antithrombin complexes are created rapidly after the production of thrombin and therefore have been identified as a subclinical marker of thrombin formation.¹⁰ Several studies have used TAT as a surrogate marker for clinical thromboembolism (deep venous thrombosis or pulmonary embolism). This is useful when sample sizes are too small or observation periods too short to use clinical thrombosis as an end point. We used this marker to test the hypothesis that CS results in an increased generation of thrombin and thus, an increased rate of activation of the coagulation cascade. Because the dogs with CS had increased levels of TAT we concluded that they had increased levels of thrombin production. We believe that this is molecular evidence of a hypercoagulable state.

The hemostatic mechanism is an intricate system of procoagulant and anticoagulant forces that interact in the production of a clot; the fibrinolytic system then operates to break down that clot. These systems have evolved to maintain blood in a fluid state under normal physiologic conditions, yet are primed to prevent blood loss at the instant of injury. Several studies of humans have shown an increase in factor VIII in CS.^2-5 This rise in factor VIII seems to correlate with the level of CS and returns to normal after treatment.^2-4 Other factors in CS that have been found to be elevated include factors II, V, VII, IX, X, XI, and XII.^2-5 We found several of the procoagulant factors, specifically factors II, V, VII, IX, X, XII and fibrinogen, to be elevated in our dog model of CS. We were, however, unable to demonstrate an increase in factor VIII levels. There are few data in the literature to support a simple increase in a single (or multiple) procoagulant factor(s) as a risk factor for thrombophilia. Because of this, we were not satisfied using the increases in procoagulant factors as the sole explanation for the hypercoagulability of CS.

Although various studies of CS in humans have shown increases in coagulation factors, studies differ in the exact factors that are elevated. Only factor VIII has been reported to be significantly elevated in most studies. This may be explained in part by small sample sizes of the various studies. The normal activity of coagulation factors in humans varies from 50% to 150%. Assuming that the normal range is set at ±2σ, then the sample size needed to detect a difference of 10% with a power of .8 is 100. An SD of this magnitude is reasonable and correlates well with the SD seen in this and other studies. Most of the studies of humans had sample sizes ranging from 9 to 30. These studies may have lacked a large enough sample size to generate the power needed to see a difference in factor levels. Because the incidence of CS in humans is 2 to 5 per 1 million, finding adequate numbers of patients at one institution is difficult. The incidence of naturally occurring CS in canines is several times higher, which is why we decided to study dogs with CS.

In the current study, factor VIII was not significantly elevated in dogs with CS. This would seem to be disconcordant with the human literature in which factor VIII is known to act as an acute-phase reactant in humans.¹¹ Humans with an acute pathologic process are likely to seek treatment promptly, whereas dogs are dependent on their owners to recognize the symptoms, which may result in the treatment onset occurring later in the course of the disease. Therefore, the dogs with CS that were enrolled in our study may have had a more chronic course of disease, allowing the factor VIII levels to decrease from an acute peak. This may explain why factor VIII was not increased to the same degree as the other procoagulant factors.

A simple increase in procoagulant factors did not seem to be a reasonable explanation for the hypercoagulability; if a second insult were present, the basis for hypercoagulability would be clearer. The second insult we postulated could be in either of 2 forms. A decrease in the fibrinolytic system could result in decreased clearance of formed clots. Decreased clot clearance would lead to a clinically important excess in clots. The other potential derangement could be a decrease in the activity of antithrombin, the endogenous system designed to protect against excess thrombin (clot) formation.

A previous study reported inhibition of fibrinolytic activity in patients with CS.¹² Decreased fibrinolytic potential has been found to correlate with thrombotic events.¹³ In another study, decreased fibrinolytic potential was secondary to increased plasminogen activator inhibitor 1 levels.¹⁴ A study that compared patients with CS with control patients found that decreased fibrinolysis was also attributed to increased levels of plasminogen activator inhibitor 1.¹² We found no reduction in the fibrinolytic capability of the dogs with CS in our study and we feel our sample size was large enough to detect one had one been present. Thus, a normal to accelerated procoagulant response with a defective fibrinolytic system would not explain the hypercoagulability.

Decreases in the naturally occurring antithrombotics have been found to relate to thromboembolism in humans.¹⁵ Antithrombin is the primary naturally occurring antithrombotic in humans and dogs. Antithrombin levels in humans with CS have not previously been reported. We found a significant decrease in antithrombin levels in the dogs with CS compared with controls. This second derangement in the coagulation cascade thus supports a hypercoagulable state. The concurrent increase in procoagulant factors and decrease in anticoagulant factors may better explain the molecular cause of the hypercoagulable state seen in dogs with CS.

Of issue in the treatment of patients with CS is the use of heparin as the prophylaxis against and treatment for thromboembolic complications. Heparin functions by accelerating the activity of antithrombin. Since antithrombin levels are decreased in dogs with CS, and likely in humans with CS, prophylaxis using heparin should be monitored and adjusted to appropriate levels.

Our study has several weaknesses. We used a dog model to allow for a larger sample size than would have been possible at our center with a human model. Although CS occurs in both and the coagulation cascades are similar in both dogs and humans, it is possible that the molecular cause of the hypercoagulability in CS is dissimilar between species. We also did not study treated dogs to see if treatment (lowering of serum cortisol levels) would result in a return to the normal balance between procoagulant and antithrombotic forces.Cushing syndrome results in a hypercoagulable state as measured by markers of subclinical thrombosis (TAT) and by clinical events (pulmonary embolus and deep venous thrombosis). The hypercoagulable state that occurs in dogs with CS may be partially explained by the elevation in procoagulant factors and the decrease in the naturally occurring anticoagulant factor, antithrombin. Thromboembolism prophylaxis in patients with CS who are using heparin should be monitored to ensure effectiveness.

Presented at the Critical Care Section of the Surgical Forum, American College of Surgeons, Orlando, Fla, October 26, 1998 and at the 72nd Annual Meeting of the Pacific Coast Surgical Association, Banff, Canada, February 19, 2001.

Corresponding author and reprints: John T. Owings, MD, Department of Surgery, University of California–Davis, Medical Center, 2315 Stockton Blvd, Room 4209, Sacramento, CA 95817 (e-mail: jtowings@ucdavis.edu).

André Campbell, MD San Francisco, Calif: The incidence of Cushing's syndrome in humans is quite rare, estimated to be between 2 and 5 per million per year. Therefore, it is difficult to study this entity in humans. The authors hypothesize that Cushing's syndrome seen in humans can be simulated and studied in dogs to better characterize the exact nature of the hypercoagulable state seen in this disorder. There is a 4-fold increase in the incidence of pulmonary embolus and deep venous thrombosis in Cushing's syndrome as well as a 4-fold increase in mortality. These facts make this an important subject to study.

Several human studies have shown increases in factor VIII in Cushing's syndrome, and that returns to normal after treatment. Other factors that have been shown to be elevated include factors II, V, VII, IX, X, XI, and XII in this disorder. The study model used 86 dogs total, 56 with Cushing's syndrome and 30 in the control group. Standard assay techniques were used to measure the factor levels. Results: they found that there were significantly higher levels of procoagulant factors II, V, VI, IX, X, XII, and fibrinogen in dogs with Cushing's syndrome when compared with controls.

The second finding was the marker for subclinical thrombosis, thrombin-antithrombin complex, was noted to be significantly increased in Cushing's syndrome. The third finding was natural antithrombotic. Antithrombin was significantly decreased in dogs with Cushing's syndrome. Other factors that were assayed showed no difference including factor VIII, XI, plasminogen, and von Willebrand factors, among others.

The cause of Cushing's syndrome can be multifactorial, including ACTH (adrenocorticotropic hormone)-dependent mechanisms and ACTH-independent etiologies. The ACTH-dependent mechanism includes pituitary adenomas or the classic Cushing's disease. The second ACTH-dependent mechanism is ectopic ACH-producing tumor. The ACTH-independent mechanisms include adrenal adenomas and adrenal carcinomas. Can you tell us what type of Cushing's syndrome you thought the dogs had? This is not clear from the manuscript.

The second question I have is why did you have 56 dogs in the Cushing's syndrome group and only 30 in the controls? Can you tell us why the 2 groups were not more evenly matched? With the unveiling this week of the human genome project with much fanfare, we find that humans are not much different from other animal species. You acknowledge that in the text of your discussion, and I think we can all agree in this room that humans and dogs are quite different, and we thank God for that. Is this a valid comparison?

Your study demonstrated that dogs with Cushing's syndrome had increased procoagulant factors and that thrombin antithrombin complex may represent the molecular evidence of a hypercoagulable state. Do you plan additional studies to block cortisol and measure these levels and the impact that they have on the results that you showed this morning?

Finally, your group has developed a nice model for a rare and interesting condition in humans. I am not certain from reading your manuscript that you have proved a causal relationship between these 2 conditions. Can you elaborate further on this issue?

Charles H. Scudamore, MD, Vancouver, British Columbia: I would like to ask the authors whether the severity of the Cushing's syndrome had any relationship to the degree of the coagulation abnormalities. Do artificial cushingoid conditions that we produce have the same risks?

Orlo H. Clark, MD, San Francisco, Calif: I had the exact same 2 questions. First, can you correlate the changes in coagulation factors with the severity of the disease, that is, the blood cortisol level? Second, even though Cushing's syndrome is caused by many different tumors, the most common cause is taking steroid hormones by mouth. Is there a correlation with the type of Cushing's syndrome, such as those associated with an increased ACTH (adrenocorticotropic hormone) due to a pituitary tumor or ectopic secretion of ACT or decreased ACTH in patients with adrenal tumors or taking steroids?

Dr Owings: What etiologies did we have for the Cushing's syndrome in these dogs? They were as varied as they are in humans. Dogs, it turns out, have a preponderance of the pituitary etiology at about 85% as do humans. In this particular study what we did not include were iatrogenic causes of Cushing's syndrome, which may address in part some of the subsequent questions. Veterinarians are much more likely to give steroids than human physicians and will keep animals on steroids for a longer period of time. That is an area we chose not to look at specifically.

Why was there a difference in the number of controls in the experimental dogs? The word experimental in this case really should be parenthetical. The reason for that is these were not experimental animals but in fact were animals being brought in by their owners for treatment of Cushing's syndrome, and the control dogs were also basically well dogs that went through a fasting period for blood draws for a variety of reasons. Our control dogs paralleled our experience with standard laboratory values in the dogs, and so we didn't carry out our controls, in part for financial reasons, to an equivalent 56 dogs, which would be what we had in the experimental group.

Is the comparison between dogs and humans valid? I guess maybe more for some of us than others. I do think that since we look at the clinical findings in Cushing's syndromes in dogs, including the complications that they suffer, and since they parallel so nicely what is seen in the human experience, then I believe that there is some validity in the comparison of the biochemical etiology of this.

With regard to causal relationship, obviously that is always an important question to ask of any prospective scientific study. What we have demonstrated here is an association and truth. We have subsequent studies ongoing to see whether treatment causes return of normal hemostatic function, and results on that are pending.

Did we find that the severity of the Cushing's syndrome was correlated with the degree of hypercoagulation seen? The answer there is that we didn't do a full logistic regression, and it would have been difficult with the numbers of dogs that we had. When looking at a power analysis to really determine whether there is a linear relationship between the degree of Cushing's abnormality and the level of, for example, thrombin formation, which would have been a handy technique to use, we didn't feel that we had adequate power to perform that analysis. I think, Dr Clark, that was the first part of your question. Again, the validity of the study I think rests in certain assumptions, and the assumptions being made based on a correlate seen between the dog and the clinical arena and a human in the clinical arena, and then being willing to carry those assumptions over to the biochemical side of the equation.

The biochemical events that we saw, which are an increase in the procoagulant factors, which we believe when combined with a decrease in the natural protective mechanisms from pathologic thrombosis, resulted in increased thrombin formation. That, I believe, is what occurs in humans. It's just that the numbers in human studies, to be able to do those studies, have been small enough that some of the things like antithrombin and TAT have not been able to be adequately studied.