Article Text

Download PDFPDF

Safety of spinal anesthesia in thrombocytopenic patients: are there lessons to be learnt from oncology?
  1. Anthony M -H Ho1,
  2. Glenio B Mizubuti1 and
  3. Adrienne K Ho2
  1. 1 Department of Anesthesiology and Perioperative Medicine, Queen’s University, Kingston, Ontario, Canada
  2. 2 Department of Oncology, The Christie NHS Foundation Trust, Manchester, UK
  1. Correspondence to Glenio B Mizubuti, Department of Anesthesiology & Perioperative Medicine, Queen’s University, Kingston, ON K7L 2V7, Canada; gleniomizubuti{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Outside of anesthesia, lumbar puncture (LP) is commonly performed for diagnostic and therapeutic purposes. Patients with leukemia and lymphoma need multiple LPs for diagnosis and intrathecal chemotherapy. Thrombocytopenia, sometimes complicated by coagulopathy, is frequently present in this patient population.1 The causes of thrombocytopenia in leukemia/lymphoma include marrow infiltration, chemotherapy and radiotherapy, hypersplenism, immune destruction, sepsis and disseminated intravascular coagulation. Patients with hematological malignancies require transfusion often and, with time, can become allo-immunized. As such, oncologists are reluctant to expose their patients to blood products unless absolutely necessary and are willing to perform LP at thrombocytopenic levels sometimes considered alarming to anesthesiologists. In comparison, anesthesiologists usually can avoid spinal anesthesia (SA) in thrombocytopenia. Even if a strong case for SA can be made, the need for surgical hemostasis provides an easy justification for platelet transfusion. SA in the presence of serious thrombocytopenia is therefore a rare occurrence. There are, however, operative procedures (eg, nerve and tendon repair, ankle fracture, external cephalic version for breech presentation, uncomplicated vaginal delivery) and pain conditions that do not require ideal clotting and/or are associated with minimal blood loss or transfusion requirement. In such situations, there is no surgical pretext for platelet transfusion in a thrombocytopenic patient. The lowest platelet count at which neuraxial, or any regional block for that matter, can be ‘safely’ performed is always of considerable interest to anesthesiologists. Despite the disparity in disease acuity, therapeutic goals and risk–benefit considerations between hematology-oncology and anesthesia settings, there may be lessons to learn on this important issue from the experience of oncologists.


All authors (AMHH, GBM and AKH) independently searched the Medline database between 1946 and December 2017 using the terms thrombocytopenia, lumbar puncture, leukemia, lymphoma and spinal hematoma (SH). There was no restriction on eligibility criteria in terms of the type of study or language. The references of retrieved publications were reviewed for relevant articles. The authors then compared the results of their independent searches and finalized the list of publications to include in this analysis. Retrieved articles were not assigned quality judgments. Where appropriate, authors were contacted for raw data. There were no randomized controlled trials found. We tabulated the outcome data from case series but did not include single case reports in our risk analysis and discussion. Other series on SH not related to LP in oncological patients were not included. Data were broken down to various degrees of thrombocytopenia. In some reports, a very low platelet count triggered a platelet transfusion immediately prior to LP but the post-transfusion platelet count was not measured or reported. (From discussions with oncologists, LP is performed immediately after platelet transfusion in cases of extreme thrombocytopenia, without performing a post-transfusion platelet count.) Because of the high incidence of platelet refractoriness, we assumed that the post-transfusion platelet count in extreme thrombocytopenia would be raised to the 50–100×109/L range, and not higher. The upper limits of a one-sided 95% confidence interval (CI) based on Hanley and Lippman-Hand’s risk derivation2 are presented. A less conservative risk estimate, proposed by Quigley et al to provide a ‘more realistic’ risk through mathematically minimizing the maximum expected error in future trials of equal size,3 is also included for reference.


Our search netted eight case series describing LP in thrombocytopenic patients.4–11 A total of 13 975 LPs in patients with varying degrees of thrombocytopenia were performed and are listed in table 1. Despite significant numbers of bloody taps, no case of clinically apparent SH was found in the eight case series. Based on those numbers, the upper limit of the one-sided 95% CI of the risk of SH and the ‘best’ estimate (by Quigley et al 3) are listed in table 2.

Table 1

A summary of eight case series of lumbar puncture (LP) in oncology patients with varying degrees of thrombocytopenia

Table 2

Risk of spinal hematoma when no such complication has occurred is calculated according to the number of lumbar punctures (LPs) in each category of thrombocytopenia


The recommended threshold for epidural analgesia in the obstetric literature could serve as reference for gauging clinicians’ risk tolerance of SH associated with neuraxial anesthesia. The American College of Obstetrics and Gynecology has opined that neuraxial techniques are acceptable in parturients with platelet counts above 75–80×109/L.12 In Chestnut’s Obstetric Anesthesia, it is quoted that a minimum platelet count of 80×109/L is usually safe, and that in select circumstances, platelet count between 50 and 80×109/L may still allow neuraxial block.13 There is limited data on epidural catheterization in parturients with platelet count <50 ×109/L.14 15 Since a smaller needle is used and there is no catheterization of the epidural space, the risk of SH after SA is considered lower than that after epidural catheterization.

Hematologists-oncologists and anesthesiologists are both appropriately concerned about SH in patients with thrombocytopenia. In addition, oncologists are wary that a traumatic tap can spill potentially cancerous cells into the central nervous system during an LP. Even so, their platelet count threshold for LP is substantially below that used for neuraxial anesthesia. The American Association of Blood Banks suggests a threshold of 50×109/L for adults,16 while the latest Pediatric Oncology Group of Ontario, Canada, no doubt influenced by the safety record (some of which are reviewed here) and the pitfalls associated with exposure to blood products in leukemia and lymphoma, recommends a threshold of 20×109/L for children.17 Patients with leukemia/lymphoma require multiple blood product transfusions throughout the course of their chemotherapy and radiotherapy. This long-term transfusion dependency risks allo-immunization, leading to troubling immune transfusion refractoriness, a problem that affects 27% of patients in one study even in this era of leuko-reduction.18 Platelet transfusion before LP is, understandably, seldom taken lightly by oncologists, whose timeline of caring for their patients is measured in years to decades. Their willingness to push the limit has provided important data on LP at thrombocytopenia levels that few anesthesiologists would entertain SA or surgery. Our review suggests that the risk of clinically significant SH even in severe thrombocytopenia is very low in oncology patients, and the estimates derived from such data are far more reassuring than what is generally considered by anesthesiologists.14 Our oncology colleagues in Kingston have told us that their LPs in patients with platelet count barely above 50×109/L have not resulted in a single case of SH in the past two decades.

The risk–benefit/therapeutic calculus is very different between hematological oncology and most surgical patients. While the former should not be transfused unless absolutely necessary, and require LPs (to which there are no alternatives) to survive; the latter usually suffer less severe consequences to blood product transfusion, and can be managed with alternative anaesthetic techniques other than neuraxial block. Indeed, the decision to proceed with SA in a thrombocytopenic patient, or to opt for general, local, or other regional techniques, or to pre-transfuse, is a complex balance between the risks and benefits for that particular patient, the surgeon’s skills and the anesthesiologist’s risk tolerance, skills and judgment. Nevertheless, as the levels of thrombocytopenia at which SA can still be reasonably ‘safe’ remain unknown, the oncology data may help guide anesthesiologists’ decision-making in certain circumstances. Such data, for instance, should give pause to anesthesiologists who unconditionally reject SA because the platelet count is below 80–100×109/L, or to insist on platelet transfusion and a post-transfusion count before neuraxial block even for procedures or conditions that have low or no bleeding potential.

In addition to the quantity of platelets, their quality, which can be affected by disease states such as uremia, hypothermia, acidosis, etc, and by medications, should also be considered. Also important are how fast the platelet count has been falling, and coagulation function which is influenced by medications and disease states. We acknowledge that few anesthesiologists and pain physicians would contemplate neuraxial block when the platelet count is extremely low. The risks data presented for extreme thrombocytopenia and LP will nevertheless reassure that the risks associated with less severe thrombocytopenia are probably low enough to make neuraxial block worth considering.

We did not include individual case reports in our analysis because they do not by themselves assist in estimating the risk. What they do tell is that SH does occur at even normal levels of platelet count, and the risk is not zero. Indeed, spontaneous spinal hematoma in the absence of hemostatic deficiency or neuraxial instrumentation also rarely occurs.19 Thus, although the lower limits of the risk CIs derived statistically start from zero, we believe practically, one must acknowledge that there is a baseline risk of SH after SA even in patients without hemostatic defects, which is 8:260 000 (risk 0.00003; 95% CI 0.00001 to 0.00006), according to a Swedish study.20 This risk quote is higher, and thus more conservative, than the typically quoted 1:220 000 (=0.000005) risk of hematoma after SA.20

In addition to the important differences in the risk–benefit/therapeutic calculus between LP in oncology and SA in the perioperative setting mentioned earlier, there are other differences, as well as similarities, to consider. The spinal needle typically used for SA is pencil-point 25 G whereas oncologists typically use a more traumatic cutting-point 22 G needle for all patients, even small children. Cancer and surgery are both commonly associated with a hypercoagulable state. However, a bleeding diathesis is commonly associated with leukemia/lymphoma. Indeed, many patients seek medical help initially because of easy bruising and bleeding (such as melena, gum, nose or gynecological bleed). Patients with acute non-lymphoblastic leukemia may have qualitative platelet defects.1 In 16% of adults with acute lymphoblastic leukemia, disseminated intravascular coagulopathy can be shown on laboratory tests.1 Chemotherapeutic agents such as vincristine and L-asparaginase cause declines in certain clotting factors during induction whereas the adriamycins may activate fibrinolysis.1 In leukemia/lymphoma and in certain perioperative situations (eg, pre-eclampsia), there can be a rapid downward trend in the platelet count that could also increase the risk of hematoma. It is also not clear whether age changes the risk of SH. In a recent literature review on spontaneous SH, the mean age of 71 patients was 53.2 years (SD 23).19 Of the six SHs that occurred at those authors’ institutions, two adults were on aspirin, two adults were on warfarin and one adult was not on any antiplatelet or anticoagulant drug; one patient was an 11-year-old otherwise healthy girl whose diagnosis was made after she complained of headache and who recovered without intervention.19 In contrast, most of the series6–11 reported (table 1) are on children and adolescents (total n=13 685). The series by Vavricka et al 4 and Ning et al 5 are the only ones on adult patients (total n=290). In adults, platelet responsiveness seems to increase linearly with age (although little evidence exists in older (>75 years) adults).21 Furthermore, thromboelastography-based coagulation studies have demonstrated a tendency toward hypercoagulability with advancing age.22 All in all, caution must be exercised in translating the oncology LP experience to SA in the perioperative setting.

Future work might include updating and combining all published case series on spinal and epidural for all surgeries, deliveries and LPs. Epidural (single-shot and catheterization) should be included (as it is considered higher risk than SA) to improve the accuracy of the estimate while staying conservative. From the clinical and research perspective, thromboelastography should be considered in patients with worrisome thrombocytopenia in whom a neuraxial procedure is considered desirable.23



  • Competing interests None declared.