Summary. Background: Inherited factor (F)VII deficiency is the commonest of the rare bleeding disorders, with a wide set of hemorrhagic features. Other than for the severe clinical forms (for which treatment guidelines are well defined), consistent recommendations regarding perioperative replacement management do not exist for mild and asymptomatic FVII-deficient patients. Objectives: The present study aimed to evaluate the influence of bleeding history, FVII procoagulant activity levels (FVII:C) and the type of surgical procedure on the management of inherited FVII-deficient patients before surgery. Patients: One hundred and fifty-seven surgical procedures, performed without replacement therapy, in 83 unrelated FVII-deficient patients (median FVII:C = 5%, range 0.6%–35%) were analyzed. Results: The overall bleeding rate was 15.3%. We found a significant relationship between previous deep traumatic hematomas and bleeding at surgery, although relationships with previous common epistaxis, easy bruising and menorrhagia were not significant. The receiver-operating characteristic (ROC) curve analysis performed on the first 83 procedures allowed us to define a cut-off value of 7% with a sensitivity of 87% (negative predictive value: 94%). To enhance the sensitivity, and to take into account the potential variation resulting from non-standardized FVII:C measurements, we would suggest applying a threshold of 10%. Conclusion: We have proposed recommendations for the perioperative management of FVII-deficient patients based on FVII:C levels, a thorough bleeding history and the type of surgery involved. By applying these recommendations, minor procedures that risk only external or controlled hemorrhage can be performed in asymptomatic or mildly affected adults, even those with FVII:C levels below 10%.
Inherited factor (F)VII deficiency is the commonest of the rare recessive inherited coagulation disorders. It is characterized by both a broad phenotypic heterogeneity and by poor correlation between reported procoagulant activity and bleeding diathesis severity [1,2]. Symptoms are variable, ranging from severe hemorrhagic forms of the disease that involve intracranial hemorrhage and hemarthrosis, to mild forms, which involve muco-cutaneous hemorrhage and post-surgical hemorrhagic complications. However, some patients with residual procoagulant FVII activity (FVII:C) as low as 5% have no bleeding history and no history of excessive bleeding during surgery without the use of any replacement therapy [3–5]. Therefore, it is difficult to detect those patients who are at risk of bleeding and therefore treatment strategies are difficult to establish, especially for FVII-deficient patients before surgery or childbirth. According to previous studies, performed either on the general population  or on clotting factor-deficient patients [4,7], clinical history usually gives the best guidance. In the present study, we performed a retrospective analysis on the clinical course of 157 surgical procedures performed without replacement therapy in 83 unrelated FVII-deficient patients. The aims were (i) to determine the incidence of bleeding complications, (ii) to identify potential pre-procedural predictors of bleeding and (iii) to identify surgical procedures that might be carried out without the use of replacement therapy.
Materials and methods
Twenty-seven French Hemophilia treatment centres responsible for maintaining the French registry of FVII deficiency participated in this retrospective study. A total of 228 FVII-deficient patients (FVII:C < 35%) treated within the centres between 1998 and 2009 were included. We chose the 30% as a general hemostatic cut-off level . Then, we added 10% to take into account the potential variation owing to non-standardized FVII:C measurements. All patients underwent a face-to-face interview, detailed below, and were F7 genotyped by direct sequencing of the nine exons, exon–intron junctions and the promoter region of the F7 gene . Of these 228 patients, we excluded 113 from our analyzes because they had no history of surgery, 29 who had undergone surgery but only with replacement therapy, and three with poorly documented clinical histories. The remaining 83 unrelated FVII-deficient patients (FVII:C < 35%) had between them undergone 157 surgical procedures without pre-surgical replacement therapy. Written informed consent was obtained from each participant in accordance with the French law.
The physician in charge of each patient recorded the detailed clinical features and bleeding outcomes from previous surgical procedures using the same standardized and structured questionnaire conducted during a face-to-face interview with each patient. This questionnaire was specifically designed in 1998 for the French registry of FVII deficiency according to national recommendations (http://www.sfar.org/_docs/articles/art62_exampreop_l.pdf, p42-43 providing key questions derived from the Watson–Williams questionnaire ). Data included any history of: (i) bruising and/or gum bleed including any gingival hemorrhage that caused patient concern, easy bruising when more than five bruises had occurred in exposed areas, or bruises measuring larger than a coin in diameter; (ii) epistaxis meaning any nosebleed that caused a patient concern (recurrent epistaxis was defined as more than two episodes or one episode requiring hospital admission or blood transfusion); (iii) menorrhagia defined as menstrual bleeding lasting for more than a week and/or the presence of large clots (severe menorrhagia was defined as leading to anemia or requiring the use of oral contraceptives); (iv) traumatic or spontaneous deep hematoma; (v) traumatic or spontaneous hemarthrosis; (vi) intracranial or gastrointestinal hemorrhage; and (vii) bleeding after trauma or vaginal delivery.
For each patient, every previous surgical procedure was carefully recorded. The questionnaire focused on the date, site and type of the procedure, the existence of any replacement therapy before surgery and the outcomes in terms of both therapy and hemorrhage, using the operation notes where available. The data were obtained from surgical reports as often as possible or from the patients’ notification for the foremost surgical procedures. In order to investigate the impact of the type of surgery on the bleeding complication rate, procedures were classified according to (i) the intrinsic bleeding risk and (ii) the fibrinolytic activity level. Intrinsic bleeding risk was assessed by two independent anesthetists, blinded to the results, who classified the 32 surgical procedures into high and low risk, according to their own experience and interpretation of the literature . The fibrinolytic activity level was assessed according to reports in the literature into procedures involving tissues with known fibrinolytic activity [12,13]. Briefly, surgical sites with high fibrinolytic activity include: (i) the urinary tract as a result of the presence of urokinase in the urine, and (ii) tissues exposed to saliva owing to the presence of tissue plasminogen activator (mainly nasal and buccal mucosa) .
The frequency and percentage of the bleeding complications were first calculated according to surgical site for each procedure. We then compared bleeding complications to the presenting feature for the first surgical procedure (to avoid bias owing to intra-individual correlation) using the χ2 test or Fisher’s exact test (when the expected values in any of the cells of the contingency table were below 5). To evaluate whether FVII:C levels were significant predictors of bleeding outcomes, we performed receiver-operating characteristic (ROC) curve analysis and compared the corresponding area under the curve to 0.5. As the consequence of surgical bleeding is so serious, we chose a threshold that would maximize sensitivity. Finally, we also performed a sensitivity analysis excluding subjects with genotypes including at least one of the Arg304Gln or Arg79Gln FVII variants known to display different FVII:C levels depending on the species-derived thromboplastin preparation used. Statistical testing was performed at the conventional two-tailed α-level of 0.5 using sas version 9.1 (SAS Institute, Cary, North Carolina, USA).
Overall rate of bleeding complications
Eighty-three patients matched the final study criteria; 38 (46%) were men and 45 (54%) women. Of these, one group containing 73 patients had two mutated F7 alleles (34 homozygotes and 39 compound heterozygotes). The remaining group of 10 patients had a single mutated F7 allele (heterozygotes) and were also all carriers of the polymorphic A2–M2 F7 alleles known to decrease FVII:C levels (haplotype -401T/-323ins10/-122C, and R353Q, respectively) [14–16]. The overall median FVII activity was 5% (5% and 29.5% for each group, respectively) ranging from 0.6% to 35%. Together, the 83 patients underwent a total of 157 procedures. In this series, 24 procedures were complicated by bleeding, resulting in an overall risk of hemorrhage of 15.3%. Table 1 presents the bleeding frequency for procedures according to surgical site. The rate of bleeding complications was not significantly different for high- and low-risk procedures. Some surgical procedures, including appendicectomy (n = 13) and elective induced abortion (n = 10), had very low rates of bleeding whereas tonsillectomy–adenoidectomy (n = 20), abdominal hernia repair (n = 8) and circumcision (n = 5) displayed high rates of bleeding complications (20%, 25% and 40%, respectively) (Table 1). Of the 10 heterozygous patients undergoing 17 surgeries, only one dental extraction was complicated with bleeding that was easily controlled using local antifibrinolytics. When the 157 surgical procedures were classified according to their involvement of tissues with or without known fibrinolytic activity , we found no statistically significant difference between the rate of bleeding complications (respectively n = 16 (19.3%) vs. n = 8 (10.8%) P-value = 0.14), even when we excluded tooth extraction (respectively n = 6 (20.0%) vs. n = 8 (10.8%) P-value = 0.21). Classification using the intrinsic bleeding risk was also not a predictor of bleeding outcome. We observed 7 (13.2%) procedures complicated by bleeding when the site was classified as high risk and 17 (16.3%) when the site was classified as low risk (P-value = 0.70).
Table 1. Frequency of bleeding complications according to surgical site in untreated factor (F)VII-deficient patients
Bleeding complication N (%)
*P-value 0.70. †Surgical procedures were classified as high risk for hemorrhage by two independent anesthetists, according to the literature and personal experience.
Site with high fibrinolytic activity
Transurethral prostate resection†
Site with low fibrinolytic activity
Abdominal hernia repair†
Carpal tunnel repair
Predictive factors for the risk of hemorrhage
Statistical analysis was performed on the first surgical procedure for each patient to identify variables that could have predicted bleeding outcomes in this retrospective cohort. For the first surgical procedures for each of the 83 patients, 14 (17.0%) were complicated by bleeding. Thirty-four patients were symptomatic, displaying moderate or mild symptoms including provoked hematomas (5/83), epistaxis (18/83 including seven recurrent), bruising (19/83) and menorrhagia (19/28 post pubescent females including one with iron deficiency). Only one patient presented with severe hemorrhages namely recurrent spontaneous hematomas and hemarthrosis. We found no history of intracranial hemorrhage. Finally 49 patients were classified as asymptomatic (21 before puberty and 28 after puberty), 27 displayed only one symptom and 7 two or three. The number of symptoms according to FVII level is presented in Fig. 1. Focusing on the 10 heterozygous patients, two women were symptomatic (one with gum bleeds and the other with provoked hematomas and menorrhagia probably as a result of an additional effect of a decreased FIX:C level to 50%). Any previous deep traumatic hematomas, whether from major or minor insults, were associated with bleeding complications (P-value = 0.03). Bleeding risk tended to be higher in patients with recurrent epistaxis, although the differences did not reach statistical significance (P-value = 0.10). However, we found no significant relationship between menorrhagia, easy bruising or gingival hemorrhage and the risk of peri-operative hemorrhage (P-value = 0.55, P-value = 0.73, respectively). We examined results both before and after puberty because children may not have experienced any serious hemostatic challenge nor reached the age of menarche (for females) by the time of their first surgical procedure. We observed more bleeding complications among young patients, although the difference was not statistically significant. Indeed, 22.8% of the surgical procedures performed before puberty were associated with intra- or post-operative hemorrhage, whereas only 14.5% experienced bleeding complications after puberty (Table 2).
Table 2. Pre-operative evaluation of bleeding history in the 83 inherited factor (F)VII-deficient patients
No bleeding complications (%)
Bleeding complications (%)
Traumatic haematoma N (%)
Epistaxis N (%)
1 or 2 episodes
More than 2 episodes
Recurrent epistaxis N (%)
Easy bruising and gingival haemorrhage N (%)
Puberty N (%)
In women (n = 29)
Menorrhagia N (%)
Figure 2 shows the ROC curve, which demonstrates the discriminatory power of FVII:C levels in predicting intra- and post-operative bleeding, with an area under the curve (AUC) significantly higher than 0.5 (AUC = 0.69, IC95% = 0.53–0.85, P-value = 0.02) (Fig. 2). We chose a cut-off value towards the upper part of the curve to maximize sensitivity. The optimum threshold was 7.0% of FVII:C, with a sensitivity of 87% and specificity of 43%. With this cut-off value, 52 subjects were classified as positive (bleeding expected with FVII:C levels < 7.0%), among whom 13 subjects actually bled (positive predictive value : 25%) and 31 subjects were defined as negative (meaning non-bleeders with FVII:C > 7.0%), of whom only two bled [negative predictive value (NPV): 94%]. Sensitivity analysis, excluding subjects with at least one allele with the Arg304Gln or Arg79Gln variants, gave similar results with an AUC = 0.66, a sensitivity of 83%, a specificity of 41% and a NPV of 90% when we chose a threshold of 7.0%.
Finally, we examined whether hemorrhage resulting from the patient’s first surgical procedure affected the management of subsequent surgery. Even with hemorrhage during the first surgical procedure, not every affected patient later received replacement therapy for subsequent procedures. Indeed, n = 31 (46%) of the non-bleeders and n = 7 (47%) of those who did bleed during their first operation underwent subsequent surgery without receiving any FVII replacement therapy (P-value = 0.93). Interestingly, we found no link between bleeding during the first surgical procedure and bleeding complications during subsequent surgical procedures. Of the 38 subjects who underwent a second surgical procedure performed without replacement therapy, n = 3 (9.6%) of non-bleeders at the first surgery and n = 1 (14.2%) of the bleeders displayed excessive bleeding. (P-value = 1.00).
In the present study, we report a retrospective analysis on 157 surgical procedures performed on 83 unrelated FVII-deficient patients without replacement therapy, with an overall rate of bleeding complications below 15.3%. To our knowledge, this study is the largest of its kind: it involved all registered subjects with this rare deficiency undergoing an invasive procedure without replacement therapy across 27 centers over a period of more than 10 years. We aimed to define surgical sites with a high risk of hemorrhage as well as identifying clinical or biochemical pre-procedural variables that could predict the risk of bleeding during surgery for patients with inherited FVII deficiency. Finally, we suggest below therapeutic recommendations for peri-operative FVII-deficient patient management.
First, to stratify those procedures with a high risk of bleeding in patients with inherited FVII deficiency, we classified the 157 procedures according to the type and site of surgery (Table 1). In our series, four surgical procedures known to be associated with a significant risk of post-operative bleeding in spite of an adequate surgical technique [11,17] were performed without replacement therapy. One prostatectomy and one transurethral resection of the prostate for cancer were performed on two asymptomatic patients with FVII:C levels of 8% and 12%, respectively. One bowel resection and one cardiac malformation repair were performed on the same patient, who had presented with menorrhagia and traumatic hematomas but who had levels of FVII:C of 34%. None of these patients experienced excessive bleeding; however, the number of cases is too small to allow any conclusions to be drawn. Considering all 157 procedures, the rate of hemorrhagic complications was not statistically different between sites with a ‘high’ or ‘low’ risk of bleeding. Nor did we observe differences between those sites with high vs. low local fibrinolytic activity. It is difficult, then, to identify specific surgical procedures that present a greater risk to FVII-deficient patients than that faced by the general population.
According to our previous findings , we asked whether the evaluation of previous bleeds provided a reliable strategy for the preoperative management of patients with inherited FVII deficiency. There are two major issues: (i) are FVII-deficient patients with a positive bleeding history at higher risk of intra- and post-operative hemorrhage? (ii) Does the absence of a previous history of bleeding rule out the risk of subsequent peri-operative hemorrhage?
To address the first issue, it should be noted that the predictive power of a clinical history depends upon the precise questions asked . That is why our series used a standardized questionnaire that followed national recommendations. Nevertheless, as the present study is retrospective, it may suffer from selection bias, imperfect recall and incomplete case record documentation. A prospective study using a standardized bleeding score questionnaire [19,20] would be more robust. Our study considered the three categories of bleeding symptoms: severe, moderate and mild. Patients with a previous history of a severe bleeding diathesis, including intracranial hemorrhage or spontaneous hemarthroses, were considered to be at high risk of further severe hemorrhage and were therefore given replacement therapy. They were thus excluded from our study. ‘Moderate symptoms,’ including traumatic hematoma, recurrent epistaxis and a previous peri-operative bleed, were only partially associated with bleeding complications at surgery. Herein we confirm that previous deep hematomas secondary to trauma seem to predict those at high risk from further bleeds . On the other hand, a previous peri-operative hemorrhage was not associated with further bleeding at surgery. Thus, rather than the risk factor being uniquely associated with a patient’s phenotype it also depends on the type of surgery performed in that patient.
In our series, two out of the 10 ‘heterozygous’ patients were symptomatic. In the literature, only two groups have reported large cohorts of 128 and 499 heterozygous FVII-deficient subjects, respectively [21,22]. Both studies showed a significant percent of symptomatic FVII-deficient heterozygotes (33% and 19%, respectively). The main reported symptoms were hemorrhages of the mucous membranes and skin (e.g. epistaxis, easy bruising, hematoma, gum bleeding and menorrhagia). Nevertheless, the heterozygous status is questionable in our series as the 10 patients were also carriers for both allelic A2–M2 variants known to be associated with decreased FVII:C levels . These findings are in accordance with the low FVII:C levels ranging from 20% to 35% in this group. In the international registry of FVII deficiency, asymptomatic FVII-deficient heterozygotes had significantly lower FVII:C median levels when they were also M2 carriers than when they were non carriers (37% and 47%, respectively) . In the same way, mild or minor symptoms, such as gingival hemorrhage or easy bruising, were not associated with a higher risk of bleeding, in accordance with the literature. Gingival hemorrhage was unable to reliably identify any coagulopathy, as a result of the high prevalence of periodontitis . Given this research, such symptoms should not be considered as signs of abnormal bleeding in a preoperative bleeding history, even in patients with inherited FVII deficiency.
The second question addressed whether or not asymptomatic FVII-deficient patients were at a higher risk of bleeding during surgery than patients with normal hemostatic function. In our series, we report the occurrence of bleeding complications during surgical procedures in eight patients who were ‘asymptomatic’ until hemostatically challenged by surgery. Five of those were children at the time of surgery (age range: 1–10 years). Indeed, children could be ‘falsely asymptomatic’ patients, as the reason that no previous bleeding history exists in pediatric patients may well be because young patients have not received a prior hemostatic challenge. The lack of a significant correlation observed between intra- and postoperative hemorrhage and puberty in our series might result from the relatively small number of subjects. The three remaining ‘asymptomatic’ adults underwent a total of 11 surgical procedures, among which six were uncomplicated high-risk procedures . Nevertheless, without a large control group, it is difficult to differentiate between FVII deficiency-related complications and the standard hazards of surgery.
As clinical evaluation is sometimes inadequate, especially for children, we wondered whether FVII:C levels could reliably predict the peri-operative bleeding risk in FVII-deficient patients. It is generally acknowledged that the FVII:C test is a poor predictor for bleeding risk in individual FVII-deficient patients . Indeed, the hemostatic potential of any individual does not depend on a single factor but results from the contribution of all the procoagulant factors and their inhibitors. The insensitivity of FVII:C tests to other physiologically relevant functions may also cause a discrepancy between the assay results and the phenotype observed . Using the ROC analysis, we found an optimum cut-off value of F VII:C for bleeding diathesis at 7.0%, with a NPV of 94%. The results were similar (NPV 90%) when analyzing the restricted group excluding those patients with Arg304Gln or Arg79Gln mutations. The FVII:C measurements were not standardized, meaning that different thromboplastins from either animal or human origins were used, potentially leading to a risk of discrepancy for some FVII variants . Thus a ‘level > 10%’ could be a good compromise. This value is in accordance with one of our previous studies, where 42 FVII-deficient patients were investigated with standardized FVII:C measurements. A cut-off value of 8% successfully differentiated between bleeders and non-to-mild bleeders in FVII-deficient patients .
Treatment guidelines already exist for severe forms of the disease, such as life-threatening hemorrhage and spontaneous hemarthrosis . However, scientifically established recommendations for peri-operative replacement therapy are lacking for mild and asymptomatic FVII-deficient patients. Although it is retrospective, our study has allowed us to suggest some recommendations. For those surgical procedures that carry either a risk of a major or disabling bleed, we would recommend using FVII:C levels of 10% as a threshold in combination with a full clinical history to determine the need for replacement therapy (Fig. 3). We would recommend that FVII-deficient patients receive replacement therapy when FVII:C levels are < 10%, whatever the clinical status, whereas patients with FVII:C levels > 30% (for which the bleeding risk might be close to that of the general population) probably do not need replacement therapy. For FVII:C values between 10% and 30%, the therapeutic decision should depend on clinical data: age (pre- or post-puberty) and clinical status. For procedures associated with a risk of only minor, external bleeds, we would suggest that the therapeutic decision depends upon the clinical history and the experience of the surgical team. If replacement therapy is not prescribed pre-operatively, then such therapy (FVII or recombinant FVIIa or FFP) must be quickly and easily available. The use of an antifibrinolytic agent, such as tranexamic acid, may have a role in preventing hemorrhage during minor surgical procedures at fibrinolytic sites, namely for tooth extractions, as suggested for FXI-deficient patients . Figure 3 depicts our current recommendations. Had these recommendations been applied to the present study’s cohort, then only one major procedure would have resulted in hemorrhage: the tonsillectomy in the 48-year-old asymptomatic male with FVII:C levels of 12%. However, it is still difficult to be sure that this bleed resulted exclusively from low FVII:C levels.
A retrospective analysis of bleeding complications after surgery performed without replacement therapy in FVII-deficient patients allowed us to distinguish between the individual risk defined by clinical history, the less accurate predictor of FVII:C levels and the risk inherent to surgery itself. Preventative FVII replacement therapy, namely recombinant FVIIa or plasma derived FVII concentrate, is an expensive treatment that is not required for all surgical procedures. We have suggested that plasma FVII:C levels greater than a 10%, combined with a negative clinical history of previous bleeds, should be sufficient to guide the treatment decision for surgery that carries a risk of a major or disabling bleed. For minor procedures, a clinical history focusing on traumatic deep hematomas and recurrent epistaxis may be sufficient. These proposals, drawn from a relatively large series of patients considering the low prevalence of severe FVII deficiency, now need further evaluation in prospective studies.
The authors gratefully acknowledge the members of the French study group of inherited FVII deficiency for collecting data: B. Arnuti (Valence), C. Barro (Grenoble), B. Bastenaire (Le Chesnay), C. Behar (Reims), E. Beltan (Pointe-à-Pitre), M.A. Bertrand (Besançon), C. Biron-Andréani (Montpellier), F. Blanc-Jouvan (Annecy), J.Y. Borg (Rouen), M.E. Briquel (Nancy), C. Caron (Lille), H. Chambost (Marseille), S. Clayssens (Toulouse), A. Durin (Lyon), V. Eschwege (Paris), R. Favier (Paris), J. Goudemand (Lille), M. Hanss (Lyon), N. Hézard (Reims), F. Hurtaud-Roux (Paris), Y. Lambrey (Beauvais), V. Le Cam-Duchez (Rouen), C. Lemonnier (Saint Pierre de la Réunion), N. Liu (Surennes), E. Malpote (Pointe-à-Pitre), I. Martin-Toutain (Paris), E. Mazoyer (Bobigny), R. Meley (Saint-Etienne), N. Montaut (Pau), R. Navarro (Montpellier), C. Négrier (Lyon), P. Nguyen (Reims), E. Perez (Saint-Pierre de la Réunion), J. Peynet (Le Chesnay), S. Pierre-Louis (Fort de France), K. Pouymayou (Marseille), V. Proulle (Paris), L. Rugeri (Lyon), P. Sie (Toulouse), N. Stieltjes (Paris), M. Trossaert (Nantes), F. Valleix-Lavenu (Vichy), E. Verdy (Paris), M.C. Vergnes (Bordeaux), S. Voisin (Toulouse, France) and B. Chatelain (Yvoir, Belgique). We are also most grateful to Professor Charles Marc Samama (Paris) and Professor Pierre Albaladejo (Grenoble) for classifying surgical procedures. This work was supported in part by the Clinical Research Unit of Montpellier CHU (Programme Hospitalier de Recherche Clinique 1997).
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.