Gemcitabine-associated thrombotic microangiopathy




Gemcitabine-associated thrombotic microangiopathy (TMA) is believed to be very rare, with an estimated incidence rate of 0.015%. Indications for gemcitabine are expanding, and comprehensive characterization of this complication is therefore important.


The authors performed a retrospective chart review of all cases with gemcitabine-associated TMA diagnosed at Partners Healthcare System (Boston, MA) between January 1997 and February 2002.


Nine patients with gemcitabine-associated TMA were identified. Diagnosis was aided by clinical and laboratory features. Renal biopsy confirmed the diagnosis in two patients. The cumulative incidence of gemcitabine-associated TMA was 0.31% (8 cases among 2586 patients) when only the 8 patients with TMA who were treated at clinics associated with the current study were considered (1 patient with a TMA syndrome was transferred from another institution). The median patient age was 53 years, and the median time to development of a TMA syndrome after the initiation of gemcitabine was 8 months (range, 3–18 months), with a cumulative dose ranging from 9 to 56 g/m2. New or exacerbated hypertension was a prominent feature in 7 of 9 patients and preceded the clinical diagnosis by 0.5–10 weeks. Treatment of TMA included discontinuation of gemcitabine, antihypertensive therapy, plasma exchange, and dialysis. Outcomes are known for all nine patients. Six patients remain alive, whereas three have died of disease progression. No patient died as a direct result of TMA, but two developed kidney failure requiring dialysis, and one developed chronic renal insufficiency.


In the current series, the largest single-institution study to date, the incidence of gemcitabine-associated TMA was higher than previously reported (0.31% vs. 0.015%). Seven of nine patients developed new or exacerbated hypertension, which could be a useful early identifier of patients with gemcitabine-associated TMA syndromes. Cancer 2004. © 2004 American Cancer Society.

The pathologic hallmarks of the thrombotic microangiopathies (TMAs) include vessel wall thickening, endothelial cell swelling, intraluminal platelet thrombi, and microvascular obstruction. Clinically, TMA manifests as hemolytic uremic syndrome (HUS) and thrombotic thrombocytopenic purpura (TTP). TMA syndromes are associated with enteric infections, bone marrow transplantation, certain autoimmune conditions, and certain drugs, and familial and idiopathic forms of TMA exist.1, 2

Symmers3 first described a drug-induced TMA syndrome in 1956 in a patient with primary syphilis treated with the arsenate oxophenarsine. A variety of antitumor agents and other drugs cause TMA,4 including (most prominently) mitomycin C, which is associated with a 4–15% risk, as well as bleomycin, cisplatin, and 5-fluorouracil.5–7 The nomenclature used to describe chemotherapy-related TMA varies. The terms hemolytic uremic syndrome (HUS), C-HUS (chemotherapy-HUS or cancer-HUS), thrombotic thrombocytopenic purpura (TTP), and TMA all appear in the literature. For simplicity, we will refer to the syndrome as TMA.

First approved in 1996 for the treatment of unresectable pancreatic carcinoma, gemcitabine is a nucleoside analog. Its primary toxicities include myelosuppression and mild liver function abnormalities. More recently, gemcitabine has been approved for the treatment of bladder and advanced nonsmall cell lung carcinomas, and it has promising activity against breast and ovarian tumors. Sales of the drug reached $875 million in 2002 and were expected to exceed $1 billion in 2003.8

Over the last 8 years, case reports and small case series have implicated gemcitabine in the occurrence of TMA syndromes.9 The reported incidence of gemcitabine-associated TMA in the literature is very low, with a manufacturer's estimate of 0.015% (range, 0.008–0.078%) according to adverse event reports through 1997.10 A recent case report reviewed the 26 cases that have been documented to date.9

We were surprised by the apparent frequency of patients with gemcitabine-associated TMA at our institutions (Massachusetts General Hospital, Brigham and Women's Hospital, and the Dana-Farber Cancer Institute, all in Boston, MA) and thus undertook a retrospective chart review to investigate the clinical manifestations, treatment options, and outcomes associated with this condition. We report on nine patients who developed gemcitabine-associated TMA between 1997 and 2002. The increasing number of patients treated with gemcitabine warrants a deeper understanding of this serious complication.


Patient Selection and Protection of Study Participants

The nine patients with gemcitabine-associated TMA were identified by inquiring with faculty members in the divisions of nephrology and oncology at Massachusetts General Hospital, Brigham and Women's Hospital, and the Dana-Farber Cancer Institute. After gaining the approval of the relevant institutional review boards, accompanying medical records for each patient were obtained and reviewed. Of the 11 patients identified by the inquiry, 2 were excluded from the series due to questionable diagnoses of gemcitabine-associated TMA. One of these patients had renal failure accompanied by sepsis, with increasing prothrombin time suggestive of disseminated intravascular coagulation. Another patient never had thrombocytopenia and did not undergo renal biopsy. Research data were coded to protect patient confidentiality as required by the Department of Health and Human Services Regulations for the Protection of Human Subjects. The total number of patients treated during the study period was determined by reviewing pharmacy database records for both on-label and off-label administration of gemcitabine at Partners Healthcare Systems (PHS) clinics.

Criteria for Diagnosis of Thrombotic Microangiopathy

There are no standard laboratory values that define TMA, but the clinical triad of renal failure, thrombocytopenia, and microangiopathic hemolytic anemia is considered the hallmark of TMA syndromes.11 Patients who were included in the study had biopsy-proven TMA or met the following (prospectively evaluated) criteria within the 2 weeks leading up to diagnosis: platelet counts < 120 × 109/L; creatinine levels > 1.5 mg/dL; and evidence of microangiopathic hemolytic anemia, including normal fibrinogen along with ≥ 1+ schistocytes on peripheral smear, lactate dehydrogenase levels > 1.5 × the upper limit of normal in the absence of another obvious cause, and/or low serum haptoglobin levels. Because patients receiving gemcitabine can experience thrombocytopenia due to bone marrow suppression, we chose a platelet count that was considered to be less than the lower limit of normal at our institutions (150 × 109/L) to increase stringency. Data such as age, gender, race, body surface area, malignancy type and stage, medical history (including medications received), creatinine levels and platelet counts at the start of gemcitabine therapy, total dose of gemcitabine, time to development of TMA, blood pressure, clinical symptoms before the development of TMA, laboratory parameters at the time of diagnosis of TMA, treatments received, and outcome were systematically extracted from the clinical record. Baseline blood pressure readings before the initiation of gemcitabine therapy were calculated based on office visits over the preceding 3 years.

Patients were considered to have existing hypertension if they were receiving antihypertensive medications or had 3 documented blood pressure readings > 140 mm Hg systolic (SBP) or 90 mm Hg diastolic (DBP). Our criteria for new or exacerbated hypertension after the initiation of gemcitabine therapy included a documented SBP reading ≥ 170 mm Hg and a sustained increase (over 2 or more readings separated by at least 1 week) of > 20 mm Hg relative to baseline in either SBP or DBP. Due to our small sample size and limited number of blood pressure readings, no statistical analysis was performed.


Table 1 displays the basic demographic features and clinical characteristics of the nine patients included in the study. These patients included 5 females and 4 males with a median age of 53 years (range, 33–67 years). Tumor types included pancreatic tumors (n = 3), sarcoma (n = 2), breast tumors (n = 1), lymphoma (n = 1), hepatocellular carcinoma (n = 1), and pseudomyxoma peritonei (n = 1). All patients had metastatic disease at the initiation of gemcitabine therapy, and eight of nine had received previous chemotherapy. No patient had a previous episode of TMA except for Patient 4, who experienced eclampsia during her first pregnancy, at age 19.

Table 1. Clinical Characteristics
Patient no.Age (yrs)GenderTumor typePrevious chemotherapy?Previous HTN?New/exacerbated HTN?Time to TMA diagnosis (mos)aCumulative gemcitabine dose (g/m2)Baseline Cr concentration (mg/dL)
  • F: female; M: male; HCC: hepatocellular carcinoma; HTN: hypertension; TMA: thrombotic microangiopathy; Cr: creatinine.

  • a

    From the start of gemcitabine treatment.

551FPseudomyxoma peritoneiYesYesYes10561
933FHodgkin lymphomaYesNoYes7110.9

In the current series, the median cumulative gemcitabine dose was 19.2 g/m2 (range, 9–56 g/m2), and the median time to development of a TMA syndrome was 8 months (range, 3–18 months). All patients had normal baseline renal function except for Patient 8, who received a liver transplant and was diagnosed with chronic calcineurin inhibitor toxicity (baseline plasma creatinine level, 1.6 mg/dL). All patients were diagnosed with TMA while they were still receiving gemcitabine, supporting the hypothesis that the TMA was attributable to chemotherapy and not to the underlying malignancy.5 The causality of gemcitabine in these patients was further supported by the finding that renal function improved after discontinuation of the drug in all seven patients who did not require permanent dialysis (data not shown).

Gemcitabine was administered as monotherapy for all patients except Patient 1, who received concurrent trastuzumab, and Patient 9, who received combination therapy consisting of gemcitabine, vinorelbine, and doxorubicin. The schedule of administration varied from patient to patient. Patient 7 received gemcitabine every week; Patients 2 and 9 received gemcitabine 2 out of every 3 weeks; Patients 1, 3, 4, 6, and 8 received gemcitabine 3 out of every 4 weeks; and Patient 5 received gemcitabine 7 out of every 8 weeks. All patients were experiencing disease remission at the time of TMA diagnosis except for Patient 8, who had progressive disease.

All patients in the current series met the criteria for TMA except for Patient 6, who did not have thrombocytopenia in the 2 weeks before the diagnosis was made (Table 2). She did, however, have thrombocytopenia 4 weeks before the diagnosis was made (82 × 109/L) and was diagnosed with a TMA syndrome based on the results of a renal biopsy. All nine patients had normal partial thromboplastin times, and all except for Patient 2 had normal prothrombin times (PT) at the time of diagnosis. Patient 2 had a PT of 16.9 seconds, for an international normalized ratio of 1.4. Her fibrinogen level was normal, and the slightly elevated PT may have been due to poor nutrition during a lengthy hospitalization. Fibrinogen degradation products were within normal limits in Patients 1 and 9, were mildly elevated in Patients 3 and 5 (0.5–2.0 μg/mL), and were not measured in the remaining patients. Six of eight patients had hematuria, ranging from trace to 3+, at diagnosis (data not shown). Figure 1A shows the glomerular changes in a renal biopsy tissue specimen from Patient 1 that are typical of chronic microangiopathy. Figure 1B shows the transmission electron micrograph from the same biopsy specimen, which exhibited glomerular capillary luminal narrowing, endotheliosis, and reduplication of the basement membrane.

Table 2. TMA Diagnosis Parameters
Patient no.Haptoglobin (normal, 30–200 mg/dL)LDH (normal, 110–210 U/L)Platelet count (109/L)SchistocytesRenal biopsy performed?Cr concentration at diagnosis (mg/dL)
  • TMA: thrombotic microangiopathy; ND: not determined; LDH: lactate dehydrogenase; U: units; Cr: creatinine.

  • a

    Patient 6 had platelet count of 82 × 109/L four weeks before diagnosis.

  • b

    Different normal LDH range (313–618).

1< 7543113YesYes2.5
5< 687049YesNo2.7
6< 71305237aNoYes2.0
8< 697853YesNo5.1
9< 72821b61YesNo2.1
Figure 1.

Renal biopsy findings (Patient 1) demonstrating glomerular features of chronic thrombotic microangiopathy. (A) Diffuse and marked thickening of glomerular capillary walls, with widespread basement membrane reduplication (double-contour formation) in a chainlike pattern (arrows). (B) Subendothelial basement membrane reduplication (*), with interposed cells (#) and flocculent-to-fibrous matrix material between original and reduplicated basement membrane layers. Podocytes exhibit degenerative changes, including widespread foot process effacement (>), microvillous degeneration, and vacuolization. Periodic acid–Schiff stain (A); transmission electron micrograph (B). Original magnification ×400 (A); ×10,000 (B).

The most common clinical symptom was new or exacerbated hypertension. Table 3 shows data on average baseline blood pressure, first blood pressure elevation, how long before diagnosis the first blood pressure elevation was documented, and maximum blood pressure before diagnosis. Seven of 9 patients developed new or exacerbated hypertension (SBP > 170 mm Hg) after the start of gemcitabine therapy. Of the 7 patients with new or exacerbated hypertension, 4 had blood pressure increases of > 40 mm Hg relative to their average baseline SBP. Some patients had a long gap between the initial documentation of elevated blood pressure and the diagnosis of a TMA syndrome. Patients 4, 5, and 6 received their diagnoses of TMA 6–10 weeks after the first documentation of new hypertension (Table 3). For the other patients, the first documented blood pressure elevation occurred ≤ 1 week before diagnosis. Infusion unit records, which usually include information on vital signs, could not be obtained for Patients 3, 5, and 8. Therefore, it is possible that they also experienced earlier blood pressure elevations. Notably, 5 of the 7 patients subsequently had severe hypertension before or at the time of diagnosis, with maximum blood pressure readings of 190–218/98–115. After admission, Patients 2, 5, and 8 developed congestive heart failure requiring intubation.

Table 3. New Hypertension or Exacerbation of Existing Hypertension before Diagnosis
Patient no.Baseline BP (mm Hg)Earliest documented BP elevation (mm Hg)Time before diagnosis (weeks)Maximum BP during week before diagnosis (mm Hg)
  1. NA: not applicable (patient did not develop hypertension); BP: blood pressure.

1125/72 (n = 2)170/1001200/100
3123/84 (n = 4)170/1001170/100
4116/75 (n = 3)170/1006170/100
5143/93 (n = 3)172/1089214/109
6153/82 (n = 10)178/9210218/114
8124/76 (n = 2)198/100< 1210/98
9109/73 (n = 3)190/1151190/115

Table 4 summarizes the modes of therapy used and patient outcomes. Gemcitabine was discontinued in all nine patients. Five of 9 patients underwent 5–10 sessions of plasmapheresis. Eight of nine patients received new antihypertensive medications in the weeks immediately after diagnosis. Three required dialysis, but one of these patients experienced full recovery of renal function. Through February 2002, six of nine patients remained alive; three others had disease progression and died of disease. No patient died directly as a result of gemcitabine-associated TMA.

Table 4. Treatment and Outcome Data
Patient no.Plasma exchange treatmentsNo. of additional antihypertensive medicationsOutcome
  1. ESRD: end-stage renal disease.

172ESRD requiring dialysis
260Chronic renal insufficiency
351Death due to progressive disease
402Death due to progressive disease
5103Transient dialysis required
603Renal recovery
702Renal recovery
8102ESRD requiring dialysis, death due to progressive disease
903Renal recovery

To calculate the cumulative incidence of gemcitabine-associated TMA at PHS, we determined the number of patients who were treated with gemcitabine over the study period by examining drug prescription records from each hospital's pharmacy. From January 1997 to February 2002, 2586 patients were treated at PHS under both approved indications and investigational protocols. Considering only the 8 patients who were treated at our clinics (1 patient had received gemcitabine at another institution and was transferred to a PHS institution with a TMA syndrome), we calculated a cumulative incidence of 0.31% for gemcitabine-associated TMA.


The purpose of the current retrospective study was to determine the frequency and clinical course of TMA at our institutions. We calculated a cumulative incidence of 0.31% in the study population, compared with the manufacturer's initial crude estimate of 0.015% (range, 0.008–0.078%).10 One possible explanation for this discrepancy is that only patients with more severe manifestations of gemcitabine-associated TMA were diagnosed in the original report, because the complication was not well known at the time of that report. The finding that 8 of 12 patients required dialysis and 2 of 12 patients died of TMA in the original series, compared with 3 of 9 patients who received dialysis and no patients who died as a result of gemcitabine treatment in the current series, supports the possibility that only the most severe cases were recognized.10 Improved outcomes in the current series may also reflect earlier diagnosis due to increases in physician awareness since 1987–1997. Alternatively, our patients may have been healthier at diagnosis than the patients treated between 1987 and 1997, as indications for the use of gemcitabine in initial treatment regimens have expanded. The patients in the current study were broadly similar to patients in the study conducted by Fung et al.10 in terms of age, metastatic disease, and length of administration and dose of gemcitabine, but comparisons between retrospective series are difficult to make. Since the original publication of the incidence estimate in 1999, the gemcitabine product insert has been revised to reflect an increased incidence of 0.25% for gemcitabine-associated TMA.12 This figure is similar to the cumulative incidence of 0.31% reported in the current study. Our figure may underestimate the true incidence, because we relied on physician memory, did not review all records of patients receiving the drug, and did not screen faculty members who left PHS during the study period.

We determined that hypertension or exacerbation of existing hypertension preceded the clinical TMA diagnosis in seven of the nine patients in the current study. We used the relatively high SBP cutoff of 170 mm Hg and an increase of ≥ 20 mm Hg relative to baseline to define new or exacerbated hypertension. In their original study, Fung et al.10 reported new hypertension in 7 of 12 patients but did not investigate the timing of this finding relative to diagnosis. Hypertension very often accompanies TMA syndromes,13, 14 and the severity of hypertension and associated arteriolar changes is correlated with poor outcome in some series.15, 16 Drug-induced TMA syndromes have been specifically associated with hypertension.6, 17, 18 Raife and Lager19 described six patients with radiation or chemotherapy-induced TMA, five of whom presented with hypertension. Flombaum et al.20 described three patients with gemcitabine-associated TMA, two of whom presented with hypertensive urgency; one of these patients required escalating doses of antihypertensive medications in the months before diagnosis.

Because hypertension plays a prominent role in TMA syndromes in general and in drug-associated TMA syndromes in particular, we speculate that more widespread recognition of this important association could lead to earlier diagnosis in some patients. Weekly visits to the infusion unit, where vital signs are measured, present a unique chance to detect new hypertension as it develops. In the current series, 3 patients had a significant lag (6–10 weeks) between the documentation of SBP ≥ 170 and the diagnosis of TMA. This suggests that new or exacerbated hypertension could be used as a clue in diagnosing a developing TMA syndrome earlier than it would otherwise be detected. Diagnosing TMA in patients receiving gemcitabine is particularly difficult, because anemia and thrombocytopenia may not be useful signs, as they are common side effects related to chemotherapy-associated myelotoxicity.9, 12 We believe that worsening thrombocytopenia or anemia in the setting of new or exacerbated hypertension and renal failure should trigger a search for evidence of microangiopathic hemolysis. In addition to regular blood pressure monitoring, we recommend that patients receiving gemcitabine for > 3 months undergo monthly monitoring of creatinine, schistocyte, and haptoglobin measurements.

Patients in the current series had clinical presentation and laboratory findings consistent with an indolent form of TMA. Reports in the literature have documented the occurrence of a renal-limited form of TMA, particularly in association with calcineurin inhibitor–based immunosuppression or bone marrow transplantation.21, 22 Localized renal TMA can develop in the absence of the typical hematologic disturbances,19 and cases have been described in which localized TMA was diagnosed by renal biopsy but subsequently progressed to other organs.23, 24 Phase II trials of gemcitabine showed that World Health Organization (WHO) Grade 1/2 proteinuria and microscopic hematuria occurred in 58% and 41% of patients, respectively.25 Patients also developed WHO Grade 1/2 elevated levels of blood urea nitrogen (17%) and creatinine (8%). We speculate that these subclinical toxicities may reflect transient glomerular endothelial damage caused by gemcitabine. It is possible that in a small subset of patients, particularly those with stable or expanding disease who continue to receive the drug for an extended period of time, the cumulative effect of this glomerular endothelial damage results in the development of clinically evident TMA.

The incidence of gemcitabine-associated TMA may increase in coming years. Gemcitabine is being investigated in combination with a variety of other anticancer chemotherapeutic agents. As more patients respond to these combinations, patients may be exposed to the drug for extended times, potentially increasing the chance of developing a TMA syndrome. Use with other renal toxins such as cisplatin, which is known to cause a TMA syndrome when administered as a single agent, could also increase TMA risk.

The current investigation involved retrospective and uncontrolled data and thus was subject to the typical limitations of such studies. In addition, it was not possible to conclude from the current study that earlier serologic testing for TMA based on new or exacerbated hypertension would result in either earlier diagnosis or improved outcome. A prospective study will be required to arrive at such a conclusion. Despite these limitations, our findings show that a high index of suspicion for TMA is necessary for patients undergoing treatment with gemcitabine, as suggested by others.9, 20 Our data also indicate that patients who develop new or exacerbated hypertension while receiving gemcitabine should be investigated for a developing TMA. Further study is required to determine whether such a strategy would accelerate diagnosis and reduce morbidity.


The authors thank Bruce Chabner, M.D., for his critical reading of the current article.