Pathologic Fractures in Benign Neoplasms of the Fingers

Abstract

Background: To describe the rate and predictors of pathologic fractures in benign neoplasms of the finger, as variables from prediction models for pathologic fractures of the long bones of the extremities are not necessarily applicable to the bones of the hand. Methods: In this retrospective chart review, 69 histologically confirmed neoplasms affecting the bones of the fingers, 49 phalanges and 20 metacarpals, were identified in patients presenting at 2 urban hospitals over a 24-year period. Different variables were studied as possible predictors of pathologic fractures. Results: Forty-nine percent of the tumors presented as a pathologic fracture. The small finger was independently associated with pathologic fractures compared to the other fingers. Tumors of the metacarpal bones were the least likely to fracture compared to other bones. Conclusions: Almost half of neoplasms affecting the bones of the fingers presented initially with a pathologic fracture, most commonly the small finger. Therefore, a lower threshold for surgical treatment of a bone neoplasm affecting the small finger may be reasonable.

Introduction

Enchondromas are the most common primary benign tumor of bone in the hand.^1,2 Of these, approximately 50% occur in the proximal phalanx, followed by the metacarpals and middle phalanx.³ Theoretically, pathologic fractures should be more common in the phalanx, but a substantial number of patients present with a pathologic fracture of the metacarpal.

Pathologic fractures in the hand due to primary bone tumors, benign or malignant, pose a challenge for treatment. One can opt for either waiting for fracture healing before treatment of the lesion with curettage and grafting or immediate fixation while addressing the lesion.

Establishing what patients are at risk of fracturing is challenging because most prediction models are based on studies of long bones.⁴ These models often use pain, tumor size, and location as predictors of pathologic fractures; however, it is unclear how applicable these models are for bone tumors of the hand.⁵

Our aim is to report the rate of pathologic fractures among patients with primary bone tumors of the metacarpals and phalanges and to study the potential predictors of the occurrence of pathologic fracture in this clinical setting.

Materials and Methods

This study was performed under approval from our institutional review board. A waiver of informed consent was granted for this retrospective study.

We used International Classification of Diseases, Ninth Revision codes (213.5—“Benign skeletal tumor of wrist, hand, and fingers” and 170.5—“Malignant skeletal neoplasm of wrist, hand, or fingers”) to identify potential metacarpal and phalangeal bone tumors treated at 2 urban tertiary care centers between 1992 and 2015. Using statistical software, we included all patients with one of the following words in the pathology reports: “benign,” ‘malignant “tumor,” “enchondroma,” “chondroma,” “sarcoma,” “osteosarcoma,” “giant cell,” “chondrosarcoma,” or “osteoblastoma,” including misspellings followed by excluding all patients younger than 18 years of age (n = 438). After manual verification of the medical records, we identified 128 patients with 97 phalangeal bone tumors and 34 with metacarpal bone tumors, both benign and malignant. All malignant tumors (n = 20) were excluded for a more homogeneous cohort. Forty-two tumors were excluded because there were no available radiographs, resulting in 68 patients with 69 benign bone tumors.

Demographic and clinical data were gathered from the medical records, including sex, age at time of diagnosis, race, smoking status, affected finger, fractured bone (metacarpal, proximal, middle, or distal phalanx), histological diagnosis (based on pathology reports), tumor size (based on orthogonal radiographs), and presence of a fracture. Mechanism of injury was defined as atraumatic, low-energy injury (eg, falling from standing position or reported minor injury), or high-energy injury (eg, sport injury, alteration, falling from higher position).

Radiographs

Cases were staged according to both the Modified Lodwick-Madewell⁶ and Enneking⁷ classification systems. The staging of all tumors on the radiographs was done by a hand surgeon and an orthopedic oncologist on blinded radiographs. The size of the tumor was measured in 2 orthogonal views of the affected bone and categorized as a percentage of the size of the affected bone (Figure 1).

Figure 1.

The size of the tumor as a percentage of the bone (%) is calculated by dividing the tumor volume (Ht × Wt × Dt) by the bone volume (Hb × Wb × Db). The lines indicate the height, width, and depth of the respective structures.

Note. Ht = height of the tumor; Wt = width of the tumor; Dt = depth of the tumor; Hb = height of the bone; Wb = width of the bone; Db = depth of the bone.

Subjects

Of the 68 patients we included, 44 (65%) were women. The mean age was 40 years (SD, 14; 95% confidence interval [CI]: 37-43), and most patients were Caucasian (56, 87%). Forty patients (67%) were nonsmokers. The small finger (19, 28%) and the proximal phalanges (21, 30%) were the most commonly affected locations, whereas the thumb (7, 10%) and distal phalanges (11, 16%) were the least affected (Table 1). Enchondroma represented the largest group (51, 74%; Supplemental Appendix I).

Table 1.

Patient Characteristics and Factors Associated With Pathologic Finger Fractures.

Characteristic	All phalangeal bone tumors	Pathologic fractures		P value
		No	Yes
	(N = 69)	(n = 35)	(n = 34)
Demographics
Age group, y				.70
<25	10 (15)	5 (15)	5 (15)
25-50	41 (60)	19 (56)	22 (65)
>50	17 (25)	10 (29)	7 (21)
Sex, No. (%)				.62
Men	24 (35)	11 (32)	13 (38)
Women	44 (65)	23 (68)	21 (62)
Race, No. (%)a				.47
Caucasian	56 (87)	30 (91)	26 (84)
Non-Caucasian	8 (13)	3 (9.1)	5 (16)
Smoking, No. (%)b				.43
No	40 (67)	19 (61)	21 (72)
Yes	12 (20)	6 (19)	6 (21)
Former	8 (13)	6 (19)	2 (6.9)
Tumor characteristic
Affected finger, No. (%)				.15
Thumb	7 (10)	5 (14)	2 (5.9)
Index	13 (19)	8 (23)	5 (15)
Long	18 (26)	10 (29)	8 (24)
Ring	12 (17)	7 (20)	5 (15)
Small	19 (28)	5 (14)	14 (41)
Affected bone, No. (%)				0.006
Metacarpal	20 (29)	16 (46)	4 (12)
Proximal phalanx	21 (30)	10 (29)	11 (32)
Middle phalanx	17 (25)	7 (20)	10 (29)
Distal phalanx	11 (16)	2 (5.7)	9 (26)
Tumor diagnosis, No. (%)				.71
Benign neoplasm of bone	61 (88)	30 (86)	31 (91)
Benign neoplasm of soft tissue	8 (12)	5 (14)	3 (8.8)
Radiographic evaluation
Tumor size, No. (%)c				.66
≤25%	7 (11)	4 (13)	3 (10)
26%-50%	26 (43)	13 (42)	13 (43)
51%-75%	21 (34)	12 (39)	9 (30)
>75%	7 (11)	2 (6.5)	5 (17)
Lodwick stage, No. (%)c				.64
Ia	15 (22)	9 (26)	6 (18)
Ib	19 (28)	9 (26)	10 (29)
II	23 (33)	11 (31)	12 (35)
IIIa	9 (13)	4 (11)	5 (15)
IIIb	3 (4.4)	2 (5.7)	1 (2.9)
Enneking stage, No. (%)c				.10
I	37 (54)	22 (63)	15 (44)
II	18 (26)	8 (23)	10 (29)
III	14 (20)	5 (14)	9 (26)

Note. Boldfaced value indicates statistical significance at p <0.05.

^an = 64; ^bn = 60; ^cn = 61.

Statistical analyses

The frequencies and percentages of categorical variables and mean and standard deviation (SD) of continuous variables were calculated. In bivariate analysis, the Fisher’s exact test was used for categorical variables, the Student’s t test was used for continuous variables, and the Mann-Whitney U test was used for age groups to study the association with the rate of fracture. A P < .05 was considered statistically significant. Only 1 patient (1.4% of the total cohort) had more than one affected bone. Therefore, we considered all tumors to be independent events for the analysis. To mitigate confounding, variables with P < .15 in bivariate analysis were inserted in a multivariable logistic regression model to identify factors independently associated with pathologic fracture. The area under the receiver operating characteristic curve was calculated, and a Hosmer-Lemeshow test was used to assess model fit for the multivariable logistic regression model.

Results

Of the 69 bone tumors, 34 (49%) presented as pathologic fractures (Table 1). In bivariate analyses, there were significant differences between pathologic fracture rates and the affected bones (P = .006). The Enneking stage was borderline significant (P = .10). The fracture rate in small fingers was 74%, compared with 42% in the ring finger, 44% in the long finger, 38% in the index finger, and 29% in the thumb. Looking at the specific bones, the fracture rate in the metacarpal bones was only 20%, compared with 52% in the proximal phalanges, 59% in the middle phalanges, and 82% in the distal phalanges. In our data, 5 (71%) of 7 tumor size greater than 75% of the bone presented as a fracture, although not significant (Table 1).

Multivariable logistic regression analysis showed that the small finger was independently associated with presentation as a pathologic fracture (P = .050; Table 2). Tumors affecting the small finger had 5 times the odds (95% CI = 1.0-25) of being fractured compared with the long finger. In addition, the metacarpal bones were independently associated with a lower risk of presenting as a pathologic fracture compared with the proximal phalanx (P = .038, odds ratio = 0.20; 95% CI = 0.045-0.91).

Table 2.

Factors Independently Associated With Pathologic Phalangeal Fractures (N = 69).

Characteristic	Odds ratio	Lower (95% CI)	Upper (95% CI)	P value
Digit (ref.: Long)
Thumb	1.2	0.11	13	.88
Index	1.1	0.20	6.1	.90
Ring	1.0	0.19	5.7	.96
Small	5.0	1.0	25	0.050
Phalanx (ref.: Proximal)
Metacarpal	0.20	0.045	0.91	0.038
Middle	1.1	0.25	4.9	.90
Distal	3.9	0.56	27	.17
Enneking stage	1.3	0.95	1.7	.11

Note. Boldfaced value indicates statistical significance at p <0.05. Area under the receiver operating characteristic curve = 0.81; P value for Hosmer-Lemeshow test (goodness-of-fit test) = .48. CI = confidence interval.

The mechanism of injury was reported in 29 (85%) of 34 fractures. Patients reported that small finger fractures were caused by a low-energy injury (reported as small trauma, eg, while unbuckling a ski boot) or a reported atraumatic fracture in 63% of the cases (Figure 2). On the contrary, fractured metacarpal bones were caused by high-energy trauma (ski accident and door hit hand) in 2 of 3 metacarpal tumors (Figure 3).

Figure 2.

Frequencies of the causes of pathologic fractures per finger.

Figure 3.

Frequencies of the causes of pathologic fractures per bone.

In 1 patient, more than 1 bone tumor was identified. An 18-year-old white woman was diagnosed with a pathologic fracture of the metacarpal of her long finger, caused by an enchondroma. After imaging, an enchondroma was also seen in the proximal phalanx of the same finger.

Discussion

In our cohort, 49% of bone tumors in the hand presented as pathologic fractures. This rate is substantially higher than other locations.^8-10 Contrary to the Mirels’ system, size of the tumor and presence of pain were not significant predictors of the occurrence of a pathologic fracture.⁴ We only found location to be a relevant predictor. The small finger more frequently presented with a pathologic fracture compared with the other digits. In addition, bone tumors of the metacarpal were less likely to present as a pathologic fracture.

These results should be interpreted in light of the strengths and limitations of the study. As with any database search, the validity depends on coding accuracy. First, cause of fracture was not recorded in the medical records in all cases. Furthermore, the mechanism of fracture can be of unclear injury. We only classified a “low-energy” or “atraumatic” injury if this was explicitly noted in the record. Finally, it is possible that clinicians may see fractured enchondromas more frequently, but do not associate diagnosis codes with them when they are asymptomatic.

As in prior studies, we confirmed the findings that enchondromas are the most common tumors in the hand.^11,12 A similar rate (40%; 41 of 102) of enchondromas in the hand presented as pathologic fractures was found by Sassoon et al.¹³ Shenoy et al¹² reported a rate of only 22% (53 of 233) of pathological fractures as the first presenting complaint in hand tumors. Similar to our cohort, they found that the proximal phalanx was the most common bone affected by a fracture, and most of the pathologic fractures affected the fifth ray (43.9%). Riester et al¹⁴ reported a high fracture rate (95%) in the small finger affected by enchondromas and lower rate (40%) in the metacarpal. These data identify an association between pathologic fracture and involvement of the small finger. Possible explanation for the higher fracture rate may be the smaller size of the bone compared with the other digit; a similar tumor size involves a relatively greater size of the bone in the small finger. Furthermore, the ulnar position of the small finger makes it more vulnerable to injury. We observed that small fingers tended to present after relatively minimal trauma, and metacarpal fractures presented after more substantial trauma. However, it is important to recognize that this study does not establish causality.

Metacarpal bones had lower odds for a pathologic fracture, probably because of the mechanical support from bordering structures (eg, phalangeal tendons, interdigital ligaments, and muscles).¹⁵ Although not statistically significant, a lower fracture rate was seen in tumors affecting the thumb. The peripheral location of the thumb should make it more vulnerable to injuries. However, differences in tendon attachments, lack of the middle phalanx, and a relatively broader bone make it more stable.^16,17

Pathologic fractures were likely to occur when the tumor invaded the phalanges or metacarpal by more than 75%. This is in line with a previous study that reported on unicameral bone cysts in 75 children. It was found that pathologic fractures occurred when 85% of the bone is occupied by the tumor. In the hand, a lower percentage seems to be needed, potentially due to a thinner cortex compared with the long bones evaluated in the previous study.¹⁸

In conclusion, the small finger is more likely to present as a pathologic fracture, and we observed that many of these presented with minor or no trauma. In patients with a fracture, oncologic treatment generally starts 6 to 8 weeks after treatment of the fracture. The presence of a fracture makes it more difficult to create a bone window to perform curettage. This may destabilize a preexisting fracture and requires caution. When an incidental benign bone tumor of the small finger is identified, it may be reasonable to consider earlier intervention to avoid later fracture, especially in periarticular tumors.