A comparison of complication rates with other studies can be done most objectively when consideration is limited to complications involving bleeding. Our bleeding rates of 0.4% for major bleeding and 0.5% for minor bleeding fall toward the lower end of the ranges reported in the literature for both major (0.0-5.3%) and minor (0.0-5.9%) bleeding [2, 7]. The rates for minor bleeding in particular are difficult to interpret since these complications may be detected at ultrasonographic monitoring examinations but may be otherwise clinically inapparent. Studies suggest that minor intra- or perihepatic bleeding detectable by ultrasound occurs in about 18-20% of liver biopsies [2, 11, 12]. Ultrasonographic monitoring is, however, not routine in all centers performing liver punctures. In our department, it is routine following CB and ABM but not following FNAC. Our mortality rate of 0.0% corresponds with the very low rate of puncture-related mortality of 0.0-0.4% documented in the literature [2, 13–15].
With respect to needle gauge, our findings correspond with data reported from two studies using anesthetized pigs in which bleeding rate correlated with increasing diameter of the puncture needles [16–18]. In humans, a similar correlation was observed in only one large prospective study . In that study, needles with diameters of 1.8 mm and 1.6 mm were associated with a significantly higher complication rate than were thinner needles 1.2 mm and 0.8 mm in diameter. Analysis of the data, however, included findings from punctures of several different organs; in addition, computed tomography served as the imaging method and the overall complication rate, rather than simply assessment of bleeding complications, was studied. This significantly limits a direct comparison with the findings of the present study. By contrast, two other studies failed to observe an increased rate of bleeding complications in liver punctures using larger-gauge needles [20–22]. Because of the available data, however, these results do not appear to justify any corresponding conclusions. In one study, there is a lack of exact information on the number of punctures performed with each respective needle diameter ; in the other, the number of cases is too small to allow valid conclusions regarding the otherwise very low bleeding rate . In fact, the currently available data do not satisfactorily demonstrate the independence of bleeding rate from needle diameter. On the other hand, because of the large number of cases, clearly described needle diameters and the exclusive analysis of data obtained from biopsies of focal lesions, we consider our findings to be very reliable and to clearly demonstrate a significantly lower bleeding rate with the use of thinner needles. The impact of different puncture techniques (FNAC vs. CB) on these results is, in our opinion, negligible. Both were performed in nearly identical manner with respect to both the ultrasonographic guidance of the liver biopsies and the duration of the intrahepatic phase (5–30 seconds).
In the case of punctures of the hepatic parenchyma, bleeding rate associated with CB (needle diameter 1.2 mm) was approximately six-times higher than that reported for ABM (needle diameter 1.4 mm). This correlation has not previously been adequately explained although this had been suggested by data reported for early retrospective studies [2, 9, 23, 24].
An important consideration in the comparison of these two methods is the fact that, unlike ultrasonographically guided CB, ABM is performed with ultrasonographic support but not under real time conditions. However, real-time imaging does not appear to reduce the bleeding risk: significant bleeding usually results from arterial sources but small arteries cannot be visualized at ultrasound. By contrast, the constant ultrasonographic monitoring does reduce the risk of injury to the lung and gallbladder [2, 25]. The reason for the reduced bleeding rate with ABM, in our opinion, is the shorter intrahepatic phase of the puncture needle (ca. 1–3 seconds). Respiration-associated injuries such as tears of the liver capsule are less likely for this reason than with CB punctures, in which the intrahepatic phase in our institution lasts 5–30 seconds depending on puncture circumstances. This explanation has also been proposed by Grant et al. . For the indication of histological assessment of cirrhosis, however, CB appears to be superior to ABM [2, 26, 27].
With respect to examiner experience, a direct comparison between studies is difficult due to differing definitions regarding the respective examiners’ degree of experience. Only the multi-center study reported by Cadranel et al.  used the same breakdown of examiner experience selected for the present study. That study, however, showed a significant reduction in complication rate, both for minor complications, such as pain and feeling unwell, and for major complications in relation to increasing examiner experience. Other publications use different criteria for determining examiner experience but report similar findings [18, 21, 29–31]. In the study by Cadranel et al., examiners with more than 150 puncture procedures were considered experienced, while those with less than 15 procedures were considered inexperienced. No significant difference in complication rates was reported between these two groups .
In our collective, however, complication rate increase in relation to examiner experience and do so in a statistically significant manner. This unexpected observation can be explained by the fact that difficult, “high risk” biopsies are only performed by experienced examiners; the elevated complication rate can thus be attributed to the more difficult puncture conditions. Inexperienced examiners are initially assigned patients with secure puncture conditions and perform procedures under supervision. In our opinion, the clear assignment of each biopsy to the corresponding degree of examiner experience serves to underscore our findings and could be interpreted to indirectly support the findings of other authors that examiner inexperience is not a risk factor for increased complication rates [10, 18, 21]. This also underscores the importance of careful supervision and a gradual advancement of less experienced examiners to increasingly complex puncture situations. It should be noted, that the results were exclusively calculated bivariate and therefore other factors are not taken into account.
Our data support the conclusion of other publications that normal or only mildly reduced coagulation parameters do not prevent bleeding complications [2, 3, 7, 9, 10, 21, 22]. Overall, 88.9% (16/18) of instances of bleeding in the present study in patients with documented coagulation parameters occurred in patients whose coagulation parameters were within the normal reference range. Thus, our pre-interventional coagulation screening would not seem to have protected against bleeding events. Reports in the literature have found thromboplastin and PTT to be poor predictors of bleeding risk in association with surgical procedures , however, based on our findings, we do not believe that there is justification for dispensing with pre-interventional screening. By comparison, Steeff et al. found a significant correlation between the occurrence of a complication and a platelet count < 60,000/mm3; hence, these researchers advised against puncture biopsies in patients with low platelet counts . Patients with end-stage liver diseases may experience more significant bleeding [32, 33]. Considering decreasing PTT and increasing thromboplastin as constant variable, we were able to ascertain a decrease in bleeding complications. A similar correlation for the platelet count did not emerge in the statistical treatment of the data.
A comparison of punctures of focal lesions vs. parenchymal punctures failed to document an increase bleeding rate. This supports data published by Frieser et al. in the only study available to us that addresses this issue . The absence of correlation between complication rate and lesion size confirms findings by Ch Yu et al. , who, in his study of puncture-related bleeding rate in patients with hepatocellular carcinoma (HCC), established four size categories, finding no statistically significant differences in terms of complication rates.
In comparison with other studies [10, 28], our data did not support any sex-specific elevation in risk. Cadranel et al.  report that “pain” and “fear” assessed using a 10-point visual analog scale occurred more frequently in female patients. In our opinion, these parameters, because of their subjectivity, do not seem useful for a gender comparison. On the other hand, McGill et al.  reported an increased risk of bleeding rate in relation to the factor “sex” in combination with age, underlying malignancy and increase number of needle passes that was found to be statistically significant at multivariate logistic regression. Patients in our study aged 70 years and older showed a clearly higher rate of bleeding complications than did younger patients. An age-independent complication rate, as has been reported in a large study by Welch et al. , could not be confirmed by our data. In Welch’s study, the complication rate for patients > 80 years of age was compared with that observed for younger patients. However, analysis of the data considered all complications, not simply bleeding: hence, a direct comparison is difficult. When bleeding rates alone are considered, our findings confirm data reported by McGill et al. , which, on bivariate logistic regression, showed significantly more frequent bleeding in relation to increase patient age.